
From nobody Wed Jun  1 18:47:10 2016
Return-Path: <andy.donati@hotmail.com>
X-Original-To: nmlrg@ietfa.amsl.com
Delivered-To: nmlrg@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id DD7F712D152 for <nmlrg@ietfa.amsl.com>; Wed,  1 Jun 2016 18:47:07 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -4.127
X-Spam-Level: 
X-Spam-Status: No, score=-4.127 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FREEMAIL_FROM=0.001, RCVD_IN_DNSWL_LOW=-0.7, RCVD_IN_MSPIKE_H2=-0.001, RP_MATCHES_RCVD=-1.426, SPF_PASS=-0.001] autolearn=ham autolearn_force=no
Authentication-Results: ietfa.amsl.com (amavisd-new); dkim=pass (2048-bit key) header.d=hotmail.com
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id PX73nmQt7SLf for <nmlrg@ietfa.amsl.com>; Wed,  1 Jun 2016 18:47:05 -0700 (PDT)
Received: from COL004-OMC2S18.hotmail.com (col004-omc2s18.hotmail.com [65.55.34.92]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-SHA384 (256/256 bits)) (No client certificate requested) by ietfa.amsl.com (Postfix) with ESMTPS id 3DBDA128B44 for <nmlrg@irtf.org>; Wed,  1 Jun 2016 18:47:05 -0700 (PDT)
Received: from NAM03-BY2-obe.outbound.protection.outlook.com ([65.55.34.72]) by COL004-OMC2S18.hotmail.com over TLS secured channel with Microsoft SMTPSVC(7.5.7601.23008); Wed, 1 Jun 2016 18:47:04 -0700
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=hotmail.com; s=selector1; h=From:To:Date:Subject:Message-ID:Content-Type:MIME-Version; bh=SP3/nFSP76ym2COCjrepYNKcG9poA18hn7yg84alugs=; b=he1eTwz4Eh1ADTIE/Y1wsj7r9JoO+qoN4g0eZ+QyRE7YcQqYPEJ7wSDJLw0kBXEOA9aHLslvRnjj3GT+gG6iZOomZmKxk7ueVyOk5tRKZWtdAbtLsrDVX2RhAoEeEIxbok1AbcypRayQITHh1ybhJt9wXESfmmvY7vmGa6FORLWHHp5FdbgmI+qxzBfCgTjmnDseVNrmNpx5irXpxoq0h1QGk6Eq030X/J2yIvjDGFJQ05waK9PLcAwHCJ9fpki065P9LfwTLBh+WWor3nKcY5fsbSUssd7kh52BhMXLm6VU3fpZomEU2wuRj3LiMGeo/RZeHeBunb9ndE2ZLHULGA==
Received: from BY2NAM03FT049.eop-NAM03.prod.protection.outlook.com (10.152.84.52) by BY2NAM03HT201.eop-NAM03.prod.protection.outlook.com (10.152.85.124) with Microsoft SMTP Server (TLS) id 15.1.511.7; Thu, 2 Jun 2016 01:47:03 +0000
Received: from BN3PR1001MB1044.namprd10.prod.outlook.com (10.152.84.57) by BY2NAM03FT049.mail.protection.outlook.com (10.152.85.177) with Microsoft SMTP Server (TLS) id 15.1.511.7 via Frontend Transport; Thu, 2 Jun 2016 01:47:02 +0000
Received: from BN3PR1001MB1044.namprd10.prod.outlook.com ([10.165.74.8]) by BN3PR1001MB1044.namprd10.prod.outlook.com ([10.165.74.8]) with mapi id 15.01.0506.013; Thu, 2 Jun 2016 01:47:01 +0000
From: Andy Donati <andy.donati@hotmail.com>
To: Michele Zorzi <zorzi@ing.unife.it>, "nmlrg@irtf.org" <nmlrg@irtf.org>
Thread-Topic: [Nmlrg] recent paper on cognition-based networks
Thread-Index: AQHRtbK0OocEX/7qok2sdyTlwGvfOZ/VdZhH
Date: Thu, 2 Jun 2016 01:47:01 +0000
Message-ID: <BN3PR1001MB10447D4E837B1AEF273A49E1EA580@BN3PR1001MB1044.namprd10.prod.outlook.com>
References: <Pine.LNX.4.64.1605241349110.8991@dns.endif.man>
In-Reply-To: <Pine.LNX.4.64.1605241349110.8991@dns.endif.man>
Accept-Language: en-US
Content-Language: en-US
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
authentication-results: spf=softfail (sender IP is 25.152.84.57) smtp.mailfrom=hotmail.com; ing.unife.it; dkim=none (message not signed) header.d=none;ing.unife.it; dmarc=fail action=none header.from=hotmail.com;
received-spf: SoftFail (protection.outlook.com: domain of transitioning hotmail.com discourages use of 25.152.84.57 as permitted sender)
x-ms-exchange-messagesentrepresentingtype: 1
x-tmn: [gD33IJmgAd95pNaRKKhyR1Bq4osji8MO]
x-eopattributedmessage: 0
x-forefront-antispam-report: CIP:25.152.84.57; IPV:NLI; CTRY:GB; EFV:NLI; SFV:NSPM; SFS:(10019020)(98900003); DIR:OUT; SFP:1102; SCL:1; SRVR:BY2NAM03HT201; H:BN3PR1001MB1044.namprd10.prod.outlook.com; FPR:; SPF:SoftFail; MLV:ovrnspm; A:1; MX:1; LANG:en; 
x-ms-office365-filtering-correlation-id: 12a3748a-4791-4b99-1987-08d38a87caf3
x-microsoft-antispam: UriScan:; BCL:0; PCL:0; RULEID:(1601124038)(5061506196)(5061507196)(1603103041)(1603101137)(1601125047); SRVR:BY2NAM03HT201; 
x-exchange-antispam-report-cfa-test: BCL:0; PCL:0; RULEID:(432015012)(82015046); SRVR:BY2NAM03HT201; BCL:0; PCL:0; RULEID:; SRVR:BY2NAM03HT201; 
x-forefront-prvs: 0961DF5286
Content-Type: text/plain; charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
MIME-Version: 1.0
X-OriginatorOrg: hotmail.com
X-MS-Exchange-CrossTenant-originalarrivaltime: 02 Jun 2016 01:47:01.7331 (UTC)
X-MS-Exchange-CrossTenant-fromentityheader: Internet
X-MS-Exchange-CrossTenant-id: 84df9e7f-e9f6-40af-b435-aaaaaaaaaaaa
X-MS-Exchange-Transport-CrossTenantHeadersStamped: BY2NAM03HT201
X-OriginalArrivalTime: 02 Jun 2016 01:47:04.0635 (UTC) FILETIME=[AA1580B0:01D1BC70]
Archived-At: <http://mailarchive.ietf.org/arch/msg/nmlrg/nUILMWloZ5JoSbdLCezP4WX2XmU>
Subject: Re: [Nmlrg] recent paper on cognition-based networks
X-BeenThere: nmlrg@irtf.org
X-Mailman-Version: 2.1.17
Precedence: list
List-Id: Network Machine Learning Research Group <nmlrg.irtf.org>
List-Unsubscribe: <https://www.irtf.org/mailman/options/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=unsubscribe>
List-Archive: <https://mailarchive.ietf.org/arch/browse/nmlrg/>
List-Post: <mailto:nmlrg@irtf.org>
List-Help: <mailto:nmlrg-request@irtf.org?subject=help>
List-Subscribe: <https://www.irtf.org/mailman/listinfo/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=subscribe>
X-List-Received-Date: Thu, 02 Jun 2016 01:47:08 -0000

Michele,

Concepts on page 1516 under the heading of D. Unsupervised learning are aki=
n to the operant conditioning concept within the behavioral learning theory=
 of psychology and there may be a practical use of this concept described b=
elow.

Network operators are typically informed of alarmed raised and alarm cleare=
d conditions by their management applications typically by an unsolicited m=
essage sent  from the networking machine in the form of an SNMP Notificatio=
n (.a.k.a. Trap).

Today when a network device reports an undesirable condition in the form of=
 a trap, the monitor application translates this notification as an alarm r=
aised and the network operator has to manually address this issue in order =
to clear the alarm.  Upon resolution, a clear trap is sent by the network d=
evice to the management station which in turn removes the condition.   The =
network device knows that there is an undesirable condition in the network =
because it sends the trap to the central office.  But today the device does=
 not know how to get out of this undesirable state and it is up to the netw=
ork operator to fix the issue.

How can the operant conditioning model be employed by the network entity ai=
d network operations when undesirable conditions occur?  A list of differen=
t configuration options can be defined where each implements a different wa=
y of dealing with the issue on how to clear the undesirable condition.  Thi=
s list can be identified and enumerated in a switch case statement of let's=
 say option 1, 2, or 3.  The discriminating stimulus is the alarm raised co=
ndition and when this occurs, the routine will be pointed to these 3 define=
d options.  The very first time this is implemented the machine has no know=
ledge of what works best and in this case it will randomly select one of th=
e presented options.    Once the machine makes a choice and the condition c=
lears in a short predefined time interval, the option selected will gain a =
point, however if the condition is not cleared, a point is taken away (if t=
here are any to be taken away) and it will select a different option.  The =
extra points accumulated of a particular option increase the odds of that o=
ption being selected the next time the condition occurs.

In order to be fair to all 3 options, each need to be tried once before the=
 machine begins to make decisions on it's own.  Also, since any particular =
choice will likely involve a configuration change, there must be a long pre=
defined time interval when the configuration reverts back to its original s=
tate before the next choice can be made.   Also it is important to note tha=
t like today, alarm raised and alarm clear traps are still sent from the de=
vice, but an additional event based trap is sent informing the operator tha=
t a particular option was chosen automatically by the machine.

Regards,
Andy Donati
________________________________________
From: nmlrg <nmlrg-bounces@irtf.org> on behalf of Michele Zorzi <zorzi@ing.=
unife.it>
Sent: Tuesday, May 24, 2016 7:52 AM
To: nmlrg@irtf.org
Subject: [Nmlrg] recent paper on cognition-based networks

Hi all,

I would like to point you to a recent paper we published, which tries to
scope out some vision for cognitive networking beyond the more traditional
view of cognitive radio. Please find it on IEEExplore (it is open access)
at http://ieeexplore.ieee.org/xpl/abstractCitations.jsp?arnumber=3D7217798

I will very much welcome your comments and feeback, especially about the
practical feasibility of this approach. I will also welcome expressions of
interest to collaborate on furthering this into something more concrete.

Thanks a lot,

Michele

_______________________________________________
nmlrg mailing list
nmlrg@irtf.org
https://www.irtf.org/mailman/listinfo/nmlrg


From nobody Thu Jun  2 05:22:55 2016
Return-Path: <brian@innovationslab.net>
X-Original-To: nmlrg@ietfa.amsl.com
Delivered-To: nmlrg@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 77B9312D6D9 for <nmlrg@ietfa.amsl.com>; Thu,  2 Jun 2016 05:22:54 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.9
X-Spam-Level: 
X-Spam-Status: No, score=-1.9 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, RCVD_IN_DNSWL_NONE=-0.0001] autolearn=ham autolearn_force=no
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 4ZxlLiUXuPBW for <nmlrg@ietfa.amsl.com>; Thu,  2 Jun 2016 05:22:52 -0700 (PDT)
Received: from uillean.fuaim.com (uillean.fuaim.com [206.197.161.140]) (using TLSv1 with cipher ADH-AES256-SHA (256/256 bits)) (No client certificate requested) by ietfa.amsl.com (Postfix) with ESMTPS id E6EE512D6CF for <nmlrg@irtf.org>; Thu,  2 Jun 2016 05:22:52 -0700 (PDT)
Received: from clairseach.fuaim.com (clairseach-high.fuaim.com [206.197.161.158]) (using TLSv1 with cipher ADH-AES256-SHA (256/256 bits)) (No client certificate requested) by uillean.fuaim.com (Postfix) with ESMTP id A3D9488114 for <nmlrg@irtf.org>; Thu,  2 Jun 2016 05:22:50 -0700 (PDT)
Received: from clemson.jhuapl.edu (swifi-nat.jhuapl.edu [128.244.87.133]) (using TLSv1 with cipher DHE-RSA-AES128-SHA (128/128 bits)) (No client certificate requested) by clairseach.fuaim.com (Postfix) with ESMTP id 5CB20328081A for <nmlrg@irtf.org>; Thu,  2 Jun 2016 05:22:50 -0700 (PDT)
To: "nmlrg@irtf.org" <nmlrg@irtf.org>
From: Brian Haberman <brian@innovationslab.net>
Message-ID: <7c0fd2e9-30bb-7bd4-9c9d-406a4489b2ba@innovationslab.net>
Date: Thu, 2 Jun 2016 08:22:44 -0400
User-Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.10; rv:45.0) Gecko/20100101 Thunderbird/45.1.1
MIME-Version: 1.0
Content-Type: multipart/signed; micalg=pgp-sha256; protocol="application/pgp-signature"; boundary="hNE5LNX4c28oWTrHe31gR6IcKhR1hddcq"
Archived-At: <http://mailarchive.ietf.org/arch/msg/nmlrg/Mf_bNNyTLYVfR-liRqj98Y-37WQ>
Subject: [Nmlrg] Spectrum collaboration challenge
X-BeenThere: nmlrg@irtf.org
X-Mailman-Version: 2.1.17
Precedence: list
List-Id: Network Machine Learning Research Group <nmlrg.irtf.org>
List-Unsubscribe: <https://www.irtf.org/mailman/options/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=unsubscribe>
List-Archive: <https://mailarchive.ietf.org/arch/browse/nmlrg/>
List-Post: <mailto:nmlrg@irtf.org>
List-Help: <mailto:nmlrg-request@irtf.org?subject=help>
List-Subscribe: <https://www.irtf.org/mailman/listinfo/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=subscribe>
X-List-Received-Date: Thu, 02 Jun 2016 12:22:54 -0000

This is an OpenPGP/MIME signed message (RFC 4880 and 3156)
--hNE5LNX4c28oWTrHe31gR6IcKhR1hddcq
Content-Type: multipart/mixed; boundary="JDRRdEkCwNSBEwAuBAFAexIf8dxMsaPRS"
From: Brian Haberman <brian@innovationslab.net>
To: "nmlrg@irtf.org" <nmlrg@irtf.org>
Message-ID: <7c0fd2e9-30bb-7bd4-9c9d-406a4489b2ba@innovationslab.net>
Subject: Spectrum collaboration challenge

--JDRRdEkCwNSBEwAuBAFAexIf8dxMsaPRS
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable

All,
     The following announcement may be of interest to this community. I
know it is a bit lower in the stack, but there is a big focus on the
learning aspects.

http://www.darpa.mil/news-events/2016-03-23

Regards,
Brian


--JDRRdEkCwNSBEwAuBAFAexIf8dxMsaPRS--

--hNE5LNX4c28oWTrHe31gR6IcKhR1hddcq
Content-Type: application/pgp-signature; name="signature.asc"
Content-Description: OpenPGP digital signature
Content-Disposition: attachment; filename="signature.asc"

-----BEGIN PGP SIGNATURE-----

iQEcBAEBCAAGBQJXUCUZAAoJEBOZRqCi7goqh2cH/2m6zmaI5V3vAudnhgqL/MVs
sZB1jHxeF4bHJyV9wJ0TbIbjokNLSdNgUFWuB2ek66ZBYJoRI+2jvWuLuUioySTc
lm5/HFIfozHeto+ohfLlmKlJAZim3/1D3gCuwgOgJIvmHA/k2MAMWkyqQPWZBxzj
nUGe9QSR/N8cnwO203Y9tPDx6Z4eBy1WQ0KputsB8USqbjQbcNcEy7Yhb6p4sf6N
l8IWlFSJbR0HReFg8++f5dXwkr+S7DWUBF+lKFATrJiKuX9DZw14u5xzgaVcHtqb
1HGpEpcxuqVvzuiXPGREE7dmjgg6VLuLe+pH92kYZp0Wy/rFaiCJpmau3uBUOkY=
=itBf
-----END PGP SIGNATURE-----

--hNE5LNX4c28oWTrHe31gR6IcKhR1hddcq--


From nobody Fri Jun  3 02:44:56 2016
Return-Path: <jiangsheng@huawei.com>
X-Original-To: nmlrg@ietfa.amsl.com
Delivered-To: nmlrg@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 5F04312B038 for <nmlrg@ietfa.amsl.com>; Fri,  3 Jun 2016 02:44:54 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -5.647
X-Spam-Level: 
X-Spam-Status: No, score=-5.647 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, RCVD_IN_DNSWL_MED=-2.3, RCVD_IN_MSPIKE_H3=-0.01, RCVD_IN_MSPIKE_WL=-0.01, RP_MATCHES_RCVD=-1.426, SPF_PASS=-0.001] autolearn=ham autolearn_force=no
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id uDTmIPPfnPbe for <nmlrg@ietfa.amsl.com>; Fri,  3 Jun 2016 02:44:51 -0700 (PDT)
Received: from szxga01-in.huawei.com (szxga01-in.huawei.com [58.251.152.64]) (using TLSv1 with cipher RC4-SHA (128/128 bits)) (No client certificate requested) by ietfa.amsl.com (Postfix) with ESMTPS id 372F212B077 for <nmlrg@irtf.org>; Fri,  3 Jun 2016 02:44:49 -0700 (PDT)
Received: from 172.24.1.136 (EHLO nkgeml414-hub.china.huawei.com) ([172.24.1.136]) by szxrg01-dlp.huawei.com (MOS 4.3.7-GA FastPath queued) with ESMTP id DLS36833; Fri, 03 Jun 2016 17:44:45 +0800 (CST)
Received: from NKGEML515-MBS.china.huawei.com ([169.254.5.169]) by nkgeml414-hub.china.huawei.com ([10.98.56.75]) with mapi id 14.03.0235.001; Fri, 3 Jun 2016 17:44:41 +0800
From: Sheng Jiang <jiangsheng@huawei.com>
To: "nmlrg@irtf.org" <nmlrg@irtf.org>
Thread-Topic: New Version Notification for draft-jiang-nmlrg-traffic-machine-learning-00.txt
Thread-Index: AQHRvXnVjQugrPJPxkuTjfTCSERqi5/XeODw
Date: Fri, 3 Jun 2016 09:44:40 +0000
Message-ID: <5D36713D8A4E7348A7E10DF7437A4B927CA6736B@NKGEML515-MBS.china.huawei.com>
Accept-Language: en-GB, zh-CN, en-US
Content-Language: zh-CN
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
x-originating-ip: [10.111.99.197]
Content-Type: text/plain; charset="utf-8"
Content-Transfer-Encoding: base64
MIME-Version: 1.0
X-CFilter-Loop: Reflected
X-Mirapoint-Virus-RAPID-Raw: score=unknown(0), refid=str=0001.0A090205.5751518E.0043, ss=1, re=0.000, recu=0.000, reip=0.000,  cl=1, cld=1, fgs=0, ip=169.254.5.169, so=2013-06-18 04:22:30, dmn=2013-03-21 17:37:32
X-Mirapoint-Loop-Id: 63d3b67b46c99470057f547a07ed3efb
Archived-At: <http://mailarchive.ietf.org/arch/msg/nmlrg/drZwycQ9F1_F3u3Cbh_MgQtziRM>
Cc: Albert Cabellos <albert.cabellos@gmail.com>
Subject: [Nmlrg] FW: New Version Notification for draft-jiang-nmlrg-traffic-machine-learning-00.txt
X-BeenThere: nmlrg@irtf.org
X-Mailman-Version: 2.1.17
Precedence: list
List-Id: Network Machine Learning Research Group <nmlrg.irtf.org>
List-Unsubscribe: <https://www.irtf.org/mailman/options/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=unsubscribe>
List-Archive: <https://mailarchive.ietf.org/arch/browse/nmlrg/>
List-Post: <mailto:nmlrg@irtf.org>
List-Help: <mailto:nmlrg-request@irtf.org?subject=help>
List-Subscribe: <https://www.irtf.org/mailman/listinfo/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=subscribe>
X-List-Received-Date: Fri, 03 Jun 2016 09:44:54 -0000

SGksIE5NTGVycywNCg0KV2UgaGF2ZSBqdXN0IHN1Ym1pdHRlZCBhIG5ldyBkcmFmdCBmb3IgdGhl
IHByb3Bvc2VkIE5NTFJHLCBhcyBiZWxvdy4gVGhpcyBkb2N1bWVudCBpcyBpbiB0aGUgcHJpbWFy
eSBzdGFnZS4gSXQgY29sbGVjdHMgdGhlIHVzZSBjYXNlcyBwcmVzZW50ZWQgaW4gdGhlIHByb3Bv
c2VkIE5NTFJHIHNlc3Npb24gaW4gSUVURjk1IG1lZXRpbmcuIE1vcmUgY29udHJpYnV0aW9ucyBv
biB1c2UgY2FzZXMgYXJlIHdlbGNvbWUuDQoNCkFueSBjb21tZW50cywgcmV2aWV3LCBzdWdnZXN0
aW9uIGFuZCBkaXNjdXNzaW9uIGFyZSBhcHByZWNpYXRlZC4NCg0KQmVzdCByZWdhcmRzLA0KDQpT
aGVuZw0KDQo+LS0tLS1PcmlnaW5hbCBNZXNzYWdlLS0tLS0NCj5Gcm9tOiBpbnRlcm5ldC1kcmFm
dHNAaWV0Zi5vcmcgW21haWx0bzppbnRlcm5ldC1kcmFmdHNAaWV0Zi5vcmddDQo+U2VudDogRnJp
ZGF5LCBKdW5lIDAzLCAyMDE2IDU6MjUgUE0NCj5UbzogTGl1YmluZyAoTGVvKTsgUGFuYWdpb3Rp
cyBEZW1lc3RpY2hhczsgU2hlbmcgSmlhbmc7IFBlcmUgQmFybGV0OyBKZXJvbWUNCj5GcmFuY29p
czsgR2lvdmFuZSBDLiBNLiBNb3VyYQ0KPlN1YmplY3Q6IE5ldyBWZXJzaW9uIE5vdGlmaWNhdGlv
biBmb3INCj5kcmFmdC1qaWFuZy1ubWxyZy10cmFmZmljLW1hY2hpbmUtbGVhcm5pbmctMDAudHh0
DQo+DQo+DQo+QSBuZXcgdmVyc2lvbiBvZiBJLUQsIGRyYWZ0LWppYW5nLW5tbHJnLXRyYWZmaWMt
bWFjaGluZS1sZWFybmluZy0wMC50eHQNCj5oYXMgYmVlbiBzdWNjZXNzZnVsbHkgc3VibWl0dGVk
IGJ5IFNoZW5nIEppYW5nIGFuZCBwb3N0ZWQgdG8gdGhlDQo+SUVURiByZXBvc2l0b3J5Lg0KPg0K
Pk5hbWU6CQlkcmFmdC1qaWFuZy1ubWxyZy10cmFmZmljLW1hY2hpbmUtbGVhcm5pbmcNCj5SZXZp
c2lvbjoJMDANCj5UaXRsZToJCVVzZSBDYXNlcyBvZiBBcHBseWluZyBNYWNoaW5lIExlYXJuaW5n
IE1lY2hhbmlzbSB3aXRoDQo+TmV0d29yayBUcmFmZmljDQo+RG9jdW1lbnQgZGF0ZToJMjAxNi0w
Ni0wMw0KPkdyb3VwOgkJSW5kaXZpZHVhbCBTdWJtaXNzaW9uDQo+UGFnZXM6CQkxOA0KPlVSTDoN
Cj5odHRwczovL3d3dy5pZXRmLm9yZy9pbnRlcm5ldC1kcmFmdHMvZHJhZnQtamlhbmctbm1scmct
dHJhZmZpYy1tYWNoaW5lLWxlYXJuaQ0KPm5nLTAwLnR4dA0KPlN0YXR1czoNCj5odHRwczovL2Rh
dGF0cmFja2VyLmlldGYub3JnL2RvYy9kcmFmdC1qaWFuZy1ubWxyZy10cmFmZmljLW1hY2hpbmUt
bGVhcm5pbmcvDQo+SHRtbGl6ZWQ6DQo+aHR0cHM6Ly90b29scy5pZXRmLm9yZy9odG1sL2RyYWZ0
LWppYW5nLW5tbHJnLXRyYWZmaWMtbWFjaGluZS1sZWFybmluZy0wMA0KPg0KPg0KPkFic3RyYWN0
Og0KPiAgIFRoaXMgZG9jdW1lbnQgaW50cm9kdWNlcyBhIHNldCBvZiB1c2UgY2FzZXMgaW4gd2hp
Y2ggbWFjaGluZSBsZWFybmluZw0KPiAgIHRlY2hub2xvZ2llcyBhcmUgYXBwbGllZCB0byBuZXR3
b3JrIHRyYWZmaWMgcmVsZXZhbnQgYWN0aXZpdGllcywNCj4gICBpbmNsdWRpbmcgbWFjaGluZSBs
ZWFybmluZyBiYXNlZCB0cmFmZmljIGNsYXNzaWZpY2F0aW9uLCB0cmFmZmljDQo+ICAgbWFuYWdl
bWVudCwgZXRjLg0KPg0KPg0KPg0KPg0KPlBsZWFzZSBub3RlIHRoYXQgaXQgbWF5IHRha2UgYSBj
b3VwbGUgb2YgbWludXRlcyBmcm9tIHRoZSB0aW1lIG9mIHN1Ym1pc3Npb24NCj51bnRpbCB0aGUg
aHRtbGl6ZWQgdmVyc2lvbiBhbmQgZGlmZiBhcmUgYXZhaWxhYmxlIGF0IHRvb2xzLmlldGYub3Jn
Lg0KPg0KPlRoZSBJRVRGIFNlY3JldGFyaWF0DQoNCg==


From nobody Sat Jun 11 20:36:21 2016
Return-Path: <jiangsheng@huawei.com>
X-Original-To: nmlrg@ietfa.amsl.com
Delivered-To: nmlrg@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id C7D4E12B047 for <nmlrg@ietfa.amsl.com>; Sat, 11 Jun 2016 20:36:19 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -5.646
X-Spam-Level: 
X-Spam-Status: No, score=-5.646 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, HTML_MESSAGE=0.001, RCVD_IN_DNSWL_MED=-2.3, RCVD_IN_MSPIKE_H3=-0.01, RCVD_IN_MSPIKE_WL=-0.01, RP_MATCHES_RCVD=-1.426, SPF_PASS=-0.001] autolearn=ham autolearn_force=no
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id OWIOofJD9o2O for <nmlrg@ietfa.amsl.com>; Sat, 11 Jun 2016 20:36:17 -0700 (PDT)
Received: from szxga03-in.huawei.com (szxga03-in.huawei.com [119.145.14.66]) (using TLSv1 with cipher RC4-SHA (128/128 bits)) (No client certificate requested) by ietfa.amsl.com (Postfix) with ESMTPS id 1BC8312D15A for <nmlrg@irtf.org>; Sat, 11 Jun 2016 20:36:15 -0700 (PDT)
Received: from 172.24.1.136 (EHLO nkgeml411-hub.china.huawei.com) ([172.24.1.136]) by szxrg03-dlp.huawei.com (MOS 4.4.3-GA FastPath queued) with ESMTP id CCZ69854; Sun, 12 Jun 2016 11:36:09 +0800 (CST)
Received: from NKGEML515-MBX.china.huawei.com ([fe80::a54a:89d2:c471:ff]) by nkgeml411-hub.china.huawei.com ([10.98.56.70]) with mapi id 14.03.0235.001; Sun, 12 Jun 2016 11:36:06 +0800
From: Sheng Jiang <jiangsheng@huawei.com>
To: "nmlrg@irtf.org" <nmlrg@irtf.org>
Thread-Topic: Agenda for NMLRG #3, June 27th, Athens with EuCNC 2016
Thread-Index: AdHEW4qcuy1dLfNpTyemDUisQyD6xQ==
Date: Sun, 12 Jun 2016 03:36:06 +0000
Message-ID: <5D36713D8A4E7348A7E10DF7437A4B927CA8109D@NKGEML515-MBX.china.huawei.com>
Accept-Language: en-GB, zh-CN, en-US
Content-Language: zh-CN
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
x-originating-ip: [10.111.99.197]
Content-Type: multipart/alternative; boundary="_000_5D36713D8A4E7348A7E10DF7437A4B927CA8109DNKGEML515MBXchi_"
MIME-Version: 1.0
X-CFilter-Loop: Reflected
X-Mirapoint-Virus-RAPID-Raw: score=unknown(0), refid=str=0001.0A090203.575CD8AA.0010, ss=1, re=0.000, recu=0.000, reip=0.000,  cl=1, cld=1, fgs=0, ip=0.0.0.0, so=2013-05-26 15:14:31, dmn=2013-03-21 17:37:32
X-Mirapoint-Loop-Id: df3d7c51fb496e6aefbb51fc3660d9c2
Archived-At: <https://mailarchive.ietf.org/arch/msg/nmlrg/FcaTaAnFhAXaFR8iMNCh3jkNUNk>
Cc: Albert Cabellos <albert.cabellos@gmail.com>
Subject: [Nmlrg] Agenda for NMLRG #3, June 27th, Athens with EuCNC 2016
X-BeenThere: nmlrg@irtf.org
X-Mailman-Version: 2.1.17
Precedence: list
List-Id: Network Machine Learning Research Group <nmlrg.irtf.org>
List-Unsubscribe: <https://www.irtf.org/mailman/options/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=unsubscribe>
List-Archive: <https://mailarchive.ietf.org/arch/browse/nmlrg/>
List-Post: <mailto:nmlrg@irtf.org>
List-Help: <mailto:nmlrg-request@irtf.org?subject=help>
List-Subscribe: <https://www.irtf.org/mailman/listinfo/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=subscribe>
X-List-Received-Date: Sun, 12 Jun 2016 03:36:20 -0000

--_000_5D36713D8A4E7348A7E10DF7437A4B927CA8109DNKGEML515MBXchi_
Content-Type: text/plain; charset="utf-8"
Content-Transfer-Encoding: base64
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--_000_5D36713D8A4E7348A7E10DF7437A4B927CA8109DNKGEML515MBXchi_
Content-Type: text/html; charset="utf-8"
Content-Transfer-Encoding: base64
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--_000_5D36713D8A4E7348A7E10DF7437A4B927CA8109DNKGEML515MBXchi_--


From nobody Sun Jun 12 16:32:48 2016
Return-Path: <brian.e.carpenter@gmail.com>
X-Original-To: nmlrg@ietfa.amsl.com
Delivered-To: nmlrg@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 3D84A12D752 for <nmlrg@ietfa.amsl.com>; Sun, 12 Jun 2016 16:32:47 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.7
X-Spam-Level: 
X-Spam-Status: No, score=-2.7 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FREEMAIL_FROM=0.001, RCVD_IN_DNSWL_LOW=-0.7, SPF_PASS=-0.001] autolearn=ham autolearn_force=no
Authentication-Results: ietfa.amsl.com (amavisd-new); dkim=pass (2048-bit key) header.d=gmail.com
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id p8uyii_Mq8Rl for <nmlrg@ietfa.amsl.com>; Sun, 12 Jun 2016 16:32:45 -0700 (PDT)
Received: from mail-pf0-x234.google.com (mail-pf0-x234.google.com [IPv6:2607:f8b0:400e:c00::234]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by ietfa.amsl.com (Postfix) with ESMTPS id 713CE12D753 for <nmlrg@irtf.org>; Sun, 12 Jun 2016 16:32:45 -0700 (PDT)
Received: by mail-pf0-x234.google.com with SMTP id t190so39915576pfb.3 for <nmlrg@irtf.org>; Sun, 12 Jun 2016 16:32:45 -0700 (PDT)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20120113;  h=subject:to:references:from:organization:message-id:date:user-agent :mime-version:in-reply-to:content-transfer-encoding; bh=atpJXA4bcSexn+/K385vwJV3vAzT889ibeyDe14RCnY=; b=hdMscZhLNN9UaMvTkjZbvYNzcco1wQnAsraJzTTLTOJH5yKwsrDfVlxVmKjTM9rWDZ rtCkxeOe5qMJiOk0vB9h6/GqskvXkhbX6wmvDnIQ/w0FWGu6gkROu0/bkpl2cJs/9Gju hDcJzBCyubscPEQI2J2WCOfIB7004/+ggFoRfrluF03UdashIqSLy0XW3dgjqg6NTEHX WLBo1bI22eE6XUetRqgNMu2GylEXsetofR64He4aLo+9JddyyC/Pkvalhx2KbT7Sl64o WtitW2hMedLUr+QkESXrotKdvEYTNp11l0+FLDTw/hRCqTB6lHquaKe/114NHv0Mnq+y WWhQ==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:subject:to:references:from:organization :message-id:date:user-agent:mime-version:in-reply-to :content-transfer-encoding; bh=atpJXA4bcSexn+/K385vwJV3vAzT889ibeyDe14RCnY=; b=mOcHikJ3JOEX5eXQmx+v4hbKlU7NUC7q7YOPOzdHX266NFK9WobxdQ8GPyOHumAgVD w4eoFMO2Rfnf/zfnYM7nv/vlcT3gsXkgRvQHuZxwAhpMeOH5iKojuRjq9p2VjcH3U9zw 5FK5sahyrJNnC04AVcQRNeRU4cnRKhYsz/fjPMJPrgg+ql2TYgzk6i8r4stM8LMIO9yv /yOKigrfKvY7AHihVHNVQ2WMKs50wf7nrO9q2U4V4ddHXCjWfyWTqxfylb92DyYZz+Ur AujbT3XavnADNUv+IjWRb+cjQOo44OlRAWqVFH4i0/+C37sAolG7nss1WFsEcKBf/Pm/ M2Rw==
X-Gm-Message-State: ALyK8tK36cUxZgc9eJhF1ZO7odZcP+QC+wJ1srVVpprNDH/6OLnVhALUY4AWtKo4Hvm3EQ==
X-Received: by 10.98.86.72 with SMTP id k69mr18265617pfb.166.1465774364785; Sun, 12 Jun 2016 16:32:44 -0700 (PDT)
Received: from ?IPv6:2406:e007:5ebf:1:28cc:dc4c:9703:6781? ([2406:e007:5ebf:1:28cc:dc4c:9703:6781]) by smtp.gmail.com with ESMTPSA id et3sm32437276pad.48.2016.06.12.16.32.42 for <nmlrg@irtf.org> (version=TLSv1/SSLv3 cipher=OTHER); Sun, 12 Jun 2016 16:32:43 -0700 (PDT)
To: nmlrg@irtf.org
References: <20160603092508.1455.54979.idtracker@ietfa.amsl.com>
From: Brian E Carpenter <brian.e.carpenter@gmail.com>
Organization: University of Auckland
Message-ID: <05db1128-fde3-bd47-907a-9cc32136e3d8@gmail.com>
Date: Mon, 13 Jun 2016 11:32:41 +1200
User-Agent: Mozilla/5.0 (Windows NT 6.1; WOW64; rv:45.0) Gecko/20100101 Thunderbird/45.1.1
MIME-Version: 1.0
In-Reply-To: <20160603092508.1455.54979.idtracker@ietfa.amsl.com>
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit
Archived-At: <https://mailarchive.ietf.org/arch/msg/nmlrg/5zuQok2DURrsDBHrvNJNCIbgPjY>
Subject: Re: [Nmlrg] I-D Action: draft-jiang-nmlrg-traffic-machine-learning-00.txt
X-BeenThere: nmlrg@irtf.org
X-Mailman-Version: 2.1.17
Precedence: list
List-Id: Network Machine Learning Research Group <nmlrg.irtf.org>
List-Unsubscribe: <https://www.irtf.org/mailman/options/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=unsubscribe>
List-Archive: <https://mailarchive.ietf.org/arch/browse/nmlrg/>
List-Post: <mailto:nmlrg@irtf.org>
List-Help: <mailto:nmlrg-request@irtf.org?subject=help>
List-Subscribe: <https://www.irtf.org/mailman/listinfo/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=subscribe>
X-List-Received-Date: Sun, 12 Jun 2016 23:32:47 -0000

Hi,

Just a couple of comments on this interesting draft:

> 1.  Introduction
...
> 
>    Among many aspects of networks, the network traffic is one of the
>    most complicated managed objectives.  Its volume is rapidly growing
>    along with the Internet explosion.  It is always dynamically
>    changing.  Most network traffic flows only last a few minutes, or
>    even shorter.

That may be true today, but perhaps not in the future. Interactive flows
for audio and video may last much longer, and with RTCWEB they will become
more common. They also present unique difficulties, because a flow may be
very high data rate at one moment in one direction (A is speaking to B)
but then unpredictably the opposite (B is speaking to A). Or a video
conference might switch from presentation (A is speaking to everybody)
to Q/A mode (individuals are speaking to A). The flows will be maintained
but their parameters will change completely.

This makes the problem much more interesting - a measurement made at time T1
will not statistically predict the flow at time T2.

...
> 3.1.  Data of the Network Traffic
...
>    Data within communication protocols  The user contents are
>       encapsulated in layered communication protocols.  Many information
>       are contained within the protocol headers, for example the source
>       and destination IP addresses in the IP header, the port numbers in
>       the TCP/UDP header, etc.  

I think it is useful to mention the traffic class (DSCP) and flow label,
because these are guaranteed to be available even if the payload is
not available. Even the ECN bits might be useful, too.

...
> 5.  Security Considerations

I think that you should discuss Privacy Considerations. There is clearly
a problem, since the data one might capture for ML is exactly the data captured
by commercial or government surveillance. With the new emphasis on resisting such surveillance, the ML approach will have to
find a way to deal with this.
Anonymization of the collected data is one approach that is commonly used in
measurement studies.

Regards
   Brian Carpenter

On 03/06/2016 21:25, internet-drafts@ietf.org wrote:
> 
> A New Internet-Draft is available from the on-line Internet-Drafts directories.
> 
> 
>         Title           : Use Cases of Applying Machine Learning Mechanism with Network Traffic
>         Authors         : Sheng Jiang
>                           Bing Liu
>                           Panagiotis Demestichas
>                           Jerome Francois
>                           Giovane C. M. Moura
>                           Pere Barlet
> 	Filename        : draft-jiang-nmlrg-traffic-machine-learning-00.txt
> 	Pages           : 18
> 	Date            : 2016-06-03
> 
> Abstract:
>    This document introduces a set of use cases in which machine learning
>    technologies are applied to network traffic relevant activities,
>    including machine learning based traffic classification, traffic
>    management, etc.
> 
> 
> The IETF datatracker status page for this draft is:
> https://datatracker.ietf.org/doc/draft-jiang-nmlrg-traffic-machine-learning/
> 
> There's also a htmlized version available at:
> https://tools.ietf.org/html/draft-jiang-nmlrg-traffic-machine-learning-00
> 
> 
> Please note that it may take a couple of minutes from the time of submission
> until the htmlized version and diff are available at tools.ietf.org.
> 
> Internet-Drafts are also available by anonymous FTP at:
> ftp://ftp.ietf.org/internet-drafts/
> 
> _______________________________________________
> I-D-Announce mailing list
> I-D-Announce@ietf.org
> https://www.ietf.org/mailman/listinfo/i-d-announce
> Internet-Draft directories: http://www.ietf.org/shadow.html
> or ftp://ftp.ietf.org/ietf/1shadow-sites.txt
> 


From nobody Mon Jun 13 19:01:21 2016
Return-Path: <brian.e.carpenter@gmail.com>
X-Original-To: nmlrg@ietfa.amsl.com
Delivered-To: nmlrg@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 37CD912D54C for <nmlrg@ietfa.amsl.com>; Mon, 13 Jun 2016 19:01:20 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.7
X-Spam-Level: 
X-Spam-Status: No, score=-2.7 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FREEMAIL_FROM=0.001, RCVD_IN_DNSWL_LOW=-0.7, SPF_PASS=-0.001] autolearn=ham autolearn_force=no
Authentication-Results: ietfa.amsl.com (amavisd-new); dkim=pass (2048-bit key) header.d=gmail.com
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id h6RsFLduabdc for <nmlrg@ietfa.amsl.com>; Mon, 13 Jun 2016 19:01:18 -0700 (PDT)
Received: from mail-pa0-x22f.google.com (mail-pa0-x22f.google.com [IPv6:2607:f8b0:400e:c03::22f]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by ietfa.amsl.com (Postfix) with ESMTPS id 6E67712D62A for <nmlrg@irtf.org>; Mon, 13 Jun 2016 19:01:18 -0700 (PDT)
Received: by mail-pa0-x22f.google.com with SMTP id b13so28784856pat.0 for <nmlrg@irtf.org>; Mon, 13 Jun 2016 19:01:18 -0700 (PDT)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20120113;  h=subject:to:references:from:organization:message-id:date:user-agent :mime-version:in-reply-to:content-transfer-encoding; bh=Oou+S3CtfJNVGPGNW711bzqpNMZjPSQRsSo+pNVvMJ8=; b=C6eOnVfVDbUJdkcC/lf0vEHPhRnERiN+Q/GfZa2m4lLszd9NSWB0kjkqLK3X2CCtS/ S4+TYp9S6bgacfKi2C2/exCQvlldNPkVDeHsRCu2DaA0SOD0vfiqx64Qbcv8FIQzvFc2 ue0xaubJbtddx3Qpazsvt//f8AfAV5mtX9S1Qf6ylHOgRUvql3+2QaE3Irr3rQgUA0ks 16vEFjdzYBWBvpgekfgvS6qhGEekWtUVu66AeHWf6a2qOlAQrqwHqlTixqS6Li7Amx3K bTh0OYaZ1EtlnL6Oye+c0XhywoJUlA4n75A68Adm+SxmwnjSeXV7+8AQ+ebABqZ2egOO usEQ==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:subject:to:references:from:organization :message-id:date:user-agent:mime-version:in-reply-to :content-transfer-encoding; bh=Oou+S3CtfJNVGPGNW711bzqpNMZjPSQRsSo+pNVvMJ8=; b=BxxS5npLsYNFxIyrVY/5BGijjOeQWUXpPWP7Q8CtnTU+OWeg/y1669jSrfwYGwOfGU MiLDhOc0A6IKXFCdt2XWUys2xInjsbwabjT8L4hJsNdJOyQ3fghJDwYaj7ukBw4YYqph +lXM4RB3T/UK9Bww258/W2XfC97PPiWRdm59OT1GOs37WQmvHrn6ilx4l6Fm0Fn3/R2+ blPyEDRt18sDPUPqSoXLkCytCbynhEFyoVSjKrgcz9sDTRURF0LpZhlASWCmKAaP8/it x0RLzzRJXAQJi3VD8xXQwyOYPI6yCY1YaV+BfJP0ygmF+6TETGdOtA1Bag0RLhF2kHGp nwiw==
X-Gm-Message-State: ALyK8tJni0+TFAvew7xkGZgmuPMZKwxex/iX4laM/l9LRHsek3Aq4dn1wipsdLORc5nEtg==
X-Received: by 10.66.190.71 with SMTP id go7mr12159875pac.146.1465869677793; Mon, 13 Jun 2016 19:01:17 -0700 (PDT)
Received: from ?IPv6:2001:df0:0:2006:c0da:ac17:5f6d:8e76? ([2001:df0:0:2006:c0da:ac17:5f6d:8e76]) by smtp.gmail.com with ESMTPSA id i8sm40552963pao.26.2016.06.13.19.01.15 (version=TLSv1/SSLv3 cipher=OTHER); Mon, 13 Jun 2016 19:01:16 -0700 (PDT)
To: Michele Zorzi <zorzi@ing.unife.it>, nmlrg@irtf.org
References: <Pine.LNX.4.64.1605241349110.8991@dns.endif.man>
From: Brian E Carpenter <brian.e.carpenter@gmail.com>
Organization: University of Auckland
Message-ID: <842d9759-65f0-8ca9-a99b-ee47f3ee3922@gmail.com>
Date: Tue, 14 Jun 2016 14:01:16 +1200
User-Agent: Mozilla/5.0 (Windows NT 6.1; WOW64; rv:45.0) Gecko/20100101 Thunderbird/45.1.1
MIME-Version: 1.0
In-Reply-To: <Pine.LNX.4.64.1605241349110.8991@dns.endif.man>
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit
Archived-At: <https://mailarchive.ietf.org/arch/msg/nmlrg/FPu6H9bN_ehYGUoogtBade9yoTM>
Subject: Re: [Nmlrg] recent paper on cognition-based networks
X-BeenThere: nmlrg@irtf.org
X-Mailman-Version: 2.1.17
Precedence: list
List-Id: Network Machine Learning Research Group <nmlrg.irtf.org>
List-Unsubscribe: <https://www.irtf.org/mailman/options/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=unsubscribe>
List-Archive: <https://mailarchive.ietf.org/arch/browse/nmlrg/>
List-Post: <mailto:nmlrg@irtf.org>
List-Help: <mailto:nmlrg-request@irtf.org?subject=help>
List-Subscribe: <https://www.irtf.org/mailman/listinfo/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=subscribe>
X-List-Received-Date: Tue, 14 Jun 2016 02:01:20 -0000

Michele,

I read the paper with interest; it even cites an article by one of
my colleagues in Auckland (Aniket Mahani).

I have a practical question. You correctly observe that computing
power is now cheap enough to consider such techniques, but I'm
wondering what that really means. Can we imagine ML inside
high-end routers? Line speed firewalls? Or will it be confined
to specialised nodes that serve up ML results to network devices?

Another question is more abstract. You focus on unsupervised
learning where (if I understand correctly) the features that are
learned have no direct relationship to features that humans
consider important. In that situation, how can we translate what
we are calling "Intent" in the IETF work on autonomics into
goals for the ML-based system? Or to give a concrete example,
suppose there is an Intent statement that jitter (more precisely,
Packet Delay Variation, RFC3393) for voice streams must be less
than N milliseconds. How do we relate that to ML-derived features
in a packet stream?

Regards
   Brian Carpenter

On 24/05/2016 23:52, Michele Zorzi wrote:
> 
> Hi all,
> 
> I would like to point you to a recent paper we published, which tries to scope out some vision for cognitive networking beyond
> the more traditional view of cognitive radio. Please find it on IEEExplore (it is open access) at
> http://ieeexplore.ieee.org/xpl/abstractCitations.jsp?arnumber=7217798
> 
> I will very much welcome your comments and feeback, especially about the practical feasibility of this approach. I will also
> welcome expressions of interest to collaborate on furthering this into something more concrete.
> 
> Thanks a lot,
> 
> Michele
> 
> _______________________________________________
> nmlrg mailing list
> nmlrg@irtf.org
> https://www.irtf.org/mailman/listinfo/nmlrg
> 


From nobody Mon Jun 13 20:16:27 2016
Return-Path: <jiangsheng@huawei.com>
X-Original-To: nmlrg@ietfa.amsl.com
Delivered-To: nmlrg@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 41B3612D5AA for <nmlrg@ietfa.amsl.com>; Mon, 13 Jun 2016 20:16:26 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -5.647
X-Spam-Level: 
X-Spam-Status: No, score=-5.647 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, RCVD_IN_DNSWL_MED=-2.3, RCVD_IN_MSPIKE_H4=-0.01, RCVD_IN_MSPIKE_WL=-0.01, RP_MATCHES_RCVD=-1.426, SPF_PASS=-0.001] autolearn=ham autolearn_force=no
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id cpmWuRHQ_mNP for <nmlrg@ietfa.amsl.com>; Mon, 13 Jun 2016 20:16:23 -0700 (PDT)
Received: from szxga02-in.huawei.com (szxga02-in.huawei.com [119.145.14.65]) (using TLSv1 with cipher RC4-SHA (128/128 bits)) (No client certificate requested) by ietfa.amsl.com (Postfix) with ESMTPS id 8CC8012D586 for <nmlrg@irtf.org>; Mon, 13 Jun 2016 20:16:18 -0700 (PDT)
Received: from 172.24.1.36 (EHLO NKGEML413-HUB.china.huawei.com) ([172.24.1.36]) by szxrg02-dlp.huawei.com (MOS 4.3.7-GA FastPath queued) with ESMTP id DIP88955; Tue, 14 Jun 2016 11:16:08 +0800 (CST)
Received: from NKGEML515-MBX.china.huawei.com ([fe80::a54a:89d2:c471:ff]) by NKGEML413-HUB.china.huawei.com ([10.98.56.74]) with mapi id 14.03.0235.001; Tue, 14 Jun 2016 11:16:03 +0800
From: Sheng Jiang <jiangsheng@huawei.com>
To: Brian E Carpenter <brian.e.carpenter@gmail.com>, "nmlrg@irtf.org" <nmlrg@irtf.org>
Thread-Topic: [Nmlrg] I-D Action: draft-jiang-nmlrg-traffic-machine-learning-00.txt
Thread-Index: AQHRxQLJqOgYH7pDj0+OAUs6ntMVDp/oPADg
Date: Tue, 14 Jun 2016 03:16:03 +0000
Message-ID: <5D36713D8A4E7348A7E10DF7437A4B927CA81E50@NKGEML515-MBX.china.huawei.com>
References: <20160603092508.1455.54979.idtracker@ietfa.amsl.com> <05db1128-fde3-bd47-907a-9cc32136e3d8@gmail.com>
In-Reply-To: <05db1128-fde3-bd47-907a-9cc32136e3d8@gmail.com>
Accept-Language: en-GB, zh-CN, en-US
Content-Language: zh-CN
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
x-originating-ip: [10.111.99.197]
Content-Type: text/plain; charset="utf-8"
Content-Transfer-Encoding: base64
MIME-Version: 1.0
X-CFilter-Loop: Reflected
X-Mirapoint-Virus-RAPID-Raw: score=unknown(0), refid=str=0001.0A020206.575F76F9.009A, ss=1, re=0.000, recu=0.000, reip=0.000,  cl=1, cld=1, fgs=0, ip=0.0.0.0, so=2013-06-18 04:22:30, dmn=2013-03-21 17:37:32
X-Mirapoint-Loop-Id: c025dc753a974fe4b463465b533ee8d1
Archived-At: <https://mailarchive.ietf.org/arch/msg/nmlrg/IAdY3_MLl8L0ULNP_gGdDnXe9rY>
Subject: Re: [Nmlrg] I-D Action: draft-jiang-nmlrg-traffic-machine-learning-00.txt
X-BeenThere: nmlrg@irtf.org
X-Mailman-Version: 2.1.17
Precedence: list
List-Id: Network Machine Learning Research Group <nmlrg.irtf.org>
List-Unsubscribe: <https://www.irtf.org/mailman/options/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=unsubscribe>
List-Archive: <https://mailarchive.ietf.org/arch/browse/nmlrg/>
List-Post: <mailto:nmlrg@irtf.org>
List-Help: <mailto:nmlrg-request@irtf.org?subject=help>
List-Subscribe: <https://www.irtf.org/mailman/listinfo/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=subscribe>
X-List-Received-Date: Tue, 14 Jun 2016 03:16:26 -0000
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From nobody Tue Jun 14 17:46:56 2016
Return-Path: <jiangsheng@huawei.com>
X-Original-To: nmlrg@ietfa.amsl.com
Delivered-To: nmlrg@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 9B5D212DA75 for <nmlrg@ietfa.amsl.com>; Tue, 14 Jun 2016 17:46:54 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -5.636
X-Spam-Level: 
X-Spam-Status: No, score=-5.636 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, HTML_MESSAGE=0.001, RCVD_IN_DNSWL_MED=-2.3, RCVD_IN_MSPIKE_H3=-0.01, RCVD_IN_MSPIKE_WL=-0.01, RP_MATCHES_RCVD=-1.426, SPF_PASS=-0.001, T_KAM_HTML_FONT_INVALID=0.01] autolearn=ham autolearn_force=no
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 7Fj7l0sz3ZzA for <nmlrg@ietfa.amsl.com>; Tue, 14 Jun 2016 17:46:51 -0700 (PDT)
Received: from szxga03-in.huawei.com (szxga03-in.huawei.com [119.145.14.66]) (using TLSv1 with cipher RC4-SHA (128/128 bits)) (No client certificate requested) by ietfa.amsl.com (Postfix) with ESMTPS id 8D93E12DA6A for <nmlrg@irtf.org>; Tue, 14 Jun 2016 17:45:57 -0700 (PDT)
Received: from 172.24.1.36 (EHLO nkgeml414-hub.china.huawei.com) ([172.24.1.36]) by szxrg03-dlp.huawei.com (MOS 4.4.3-GA FastPath queued) with ESMTP id CDH02011; Wed, 15 Jun 2016 08:45:49 +0800 (CST)
Received: from NKGEML515-MBX.china.huawei.com ([fe80::a54a:89d2:c471:ff]) by nkgeml414-hub.china.huawei.com ([10.98.56.75]) with mapi id 14.03.0235.001; Wed, 15 Jun 2016 08:45:43 +0800
From: Sheng Jiang <jiangsheng@huawei.com>
To: "nmlrg@irtf.org" <nmlrg@irtf.org>
Thread-Topic: Meeting on Autonomic Networking: Towards Industry Adoption - Call for Contributions
Thread-Index: AQHRvbDLn+Oz0NQRukeuzR4Zj8TaEp/pwcZQ
Date: Wed, 15 Jun 2016 00:45:43 +0000
Message-ID: <5D36713D8A4E7348A7E10DF7437A4B927CA82521@NKGEML515-MBX.china.huawei.com>
Accept-Language: en-GB, zh-CN, en-US
Content-Language: zh-CN
X-MS-Has-Attach: yes
X-MS-TNEF-Correlator: 
x-originating-ip: [10.111.99.197]
Content-Type: multipart/mixed; boundary="_004_5D36713D8A4E7348A7E10DF7437A4B927CA82521NKGEML515MBXchi_"
MIME-Version: 1.0
X-CFilter-Loop: Reflected
X-Mirapoint-Virus-RAPID-Raw: score=unknown(0), refid=str=0001.0A020205.5760A53E.00C6, ss=3, re=0.000, recu=0.000, reip=0.000,  vtr=str, vl=0, cl=3, cld=1, fgs=0, ip=0.0.0.0, so=2013-05-26 15:14:31, dmn=2013-03-21 17:37:32
X-Mirapoint-Loop-Id: b9c8e47f108ff299e616f00a910d44e6
Archived-At: <https://mailarchive.ietf.org/arch/msg/nmlrg/kTs1SQTzqANoxPsGp8yYqhkcZFg>
Cc: Laurent Ciavaglia <laurent.ciavaglia@nokia.com>, "Michael Behringer \(mbehring\)" <mbehring@cisco.com>
Subject: [Nmlrg] FW: Meeting on Autonomic Networking: Towards Industry Adoption - Call for Contributions
X-BeenThere: nmlrg@irtf.org
X-Mailman-Version: 2.1.17
Precedence: list
List-Id: Network Machine Learning Research Group <nmlrg.irtf.org>
List-Unsubscribe: <https://www.irtf.org/mailman/options/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=unsubscribe>
List-Archive: <https://mailarchive.ietf.org/arch/browse/nmlrg/>
List-Post: <mailto:nmlrg@irtf.org>
List-Help: <mailto:nmlrg-request@irtf.org?subject=help>
List-Subscribe: <https://www.irtf.org/mailman/listinfo/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=subscribe>
X-List-Received-Date: Wed, 15 Jun 2016 00:46:55 -0000

--_004_5D36713D8A4E7348A7E10DF7437A4B927CA82521NKGEML515MBXchi_
Content-Type: multipart/alternative;
	boundary="_000_5D36713D8A4E7348A7E10DF7437A4B927CA82521NKGEML515MBXchi_"

--_000_5D36713D8A4E7348A7E10DF7437A4B927CA82521NKGEML515MBXchi_
Content-Type: text/plain; charset="utf-8"
Content-Transfer-Encoding: base64
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--_000_5D36713D8A4E7348A7E10DF7437A4B927CA82521NKGEML515MBXchi_
Content-Type: text/html; charset="utf-8"
Content-Transfer-Encoding: base64
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=

--_000_5D36713D8A4E7348A7E10DF7437A4B927CA82521NKGEML515MBXchi_--

--_004_5D36713D8A4E7348A7E10DF7437A4B927CA82521NKGEML515MBXchi_
Content-Type: application/pdf;
	name="Meeting on Autonomic Networking 2016.pdf"
Content-Description: Meeting on Autonomic Networking 2016.pdf
Content-Disposition: attachment;
	filename="Meeting on Autonomic Networking 2016.pdf"; size=265801;
	creation-date="Fri, 03 Jun 2016 15:58:39 GMT";
	modification-date="Fri, 03 Jun 2016 15:58:39 GMT"
Content-ID: <62D329EDDCF30B4F96B6F2B55E7B2189@huawei.com>
Content-Transfer-Encoding: base64
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--_004_5D36713D8A4E7348A7E10DF7437A4B927CA82521NKGEML515MBXchi_--



From nobody Sun Jun 19 08:47:41 2016
Return-Path: <pdemestichas@gmail.com>
X-Original-To: nmlrg@ietfa.amsl.com
Delivered-To: nmlrg@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 970AB12D54F for <nmlrg@ietfa.amsl.com>; Sun, 19 Jun 2016 08:47:40 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.4
X-Spam-Level: 
X-Spam-Status: No, score=-2.4 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, FREEMAIL_FORGED_FROMDOMAIN=0.198, FREEMAIL_FROM=0.001, HEADER_FROM_DIFFERENT_DOMAINS=0.001, HTML_MESSAGE=0.001, RCVD_IN_DNSWL_LOW=-0.7, SPF_PASS=-0.001] autolearn=ham autolearn_force=no
Authentication-Results: ietfa.amsl.com (amavisd-new); dkim=pass (2048-bit key) header.d=gmail.com
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id DO3qVydOUjwO for <nmlrg@ietfa.amsl.com>; Sun, 19 Jun 2016 08:47:37 -0700 (PDT)
Received: from mail-oi0-x22c.google.com (mail-oi0-x22c.google.com [IPv6:2607:f8b0:4003:c06::22c]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by ietfa.amsl.com (Postfix) with ESMTPS id A783712D53D for <nmlrg@irtf.org>; Sun, 19 Jun 2016 08:47:37 -0700 (PDT)
Received: by mail-oi0-x22c.google.com with SMTP id u201so179691302oie.0 for <nmlrg@irtf.org>; Sun, 19 Jun 2016 08:47:37 -0700 (PDT)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20120113;  h=mime-version:sender:in-reply-to:references:from:date:message-id :subject:to:cc; bh=69cRpg7D2MZcUonvHhbjIEOqwKkxI4PJYY5rMdAkmFc=; b=wzhj776CMDE61cqdyFFRMYQLOKFOsP58JziBaZa7HouTyAU/+P0m37Wh6OKJ+/03I5 s76AarSw/QcPM/wRepzEw/7pK87RD+47piMQlRCA+x3eKOhie5HmAOo2G296UBZ3931M 7y6jEU76n0JNtr+1b5eTjuw10H0H9+ZgPNp3L4OYSo6fbzeiMTZluL9izR+m4vzUDvRd DGUArfsx+E98g8+5IN1u3trDlbGM3IBhJRUYacBZiJW90qJeZq4kysXHHtZpSOBKu1ZS Do0Kx2wYOs8YFiS11N+rz4nX84JkaB3c+Ixk6sxWDajiQ52GdTgTQ3MvdkClZ21fG0FK qgjA==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:sender:in-reply-to:references:from :date:message-id:subject:to:cc; bh=69cRpg7D2MZcUonvHhbjIEOqwKkxI4PJYY5rMdAkmFc=; b=Rj77lzL/7LFS26H96KyC8JSTA8pcEdQ+Es5c4nkvrm7yNtZz+0ruWzteSE7ecvV5Hb vbvb1WW5YPdo+tOpbosk/kvnXBbCMwkDllc3p0gr9Jngjs+uGRl70/9EwmFVReIRwUsp cs1+DsXDHHfTPhOd/OPQUj+nCh8gJs0z/SHsKqtaHdoBgq61aSmVaJ+fJ5gBIuAlsP5j B7YUzLCxQX1AfVg686e3W1uwwUnLogsTEG8mT+GScp6goSfjXA2rMSep1H1KCf0N2kaQ VZPyegN+OJy9gB6MyVP5Vu00qAnmLvMR3cI5/jLWEehuUZ4VvTPHFKb9gr/exqtcQ/zl D0Nw==
X-Gm-Message-State: ALyK8tKCmxiZhrqDDGQD0ZuPh23hjpO6Zp8XuCgful+jc7/RG1GjwNfyOAM/OKA9T4BJiTqNItHcfu9hriipQg==
X-Received: by 10.202.172.65 with SMTP id v62mr6561014oie.30.1466351256718; Sun, 19 Jun 2016 08:47:36 -0700 (PDT)
MIME-Version: 1.0
Sender: pdemestichas@gmail.com
Received: by 10.157.17.7 with HTTP; Sun, 19 Jun 2016 08:47:07 -0700 (PDT)
In-Reply-To: <842d9759-65f0-8ca9-a99b-ee47f3ee3922@gmail.com>
References: <Pine.LNX.4.64.1605241349110.8991@dns.endif.man> <842d9759-65f0-8ca9-a99b-ee47f3ee3922@gmail.com>
From: Panagiotis Demestichas <pdemest@unipi.gr>
Date: Sun, 19 Jun 2016 18:47:07 +0300
X-Google-Sender-Auth: aAZmgMbqWZj_RSlTO6TcciVP6r4
Message-ID: <CAKo0eZO7Jon4+8MOjkiZzBO-8tLSoHvpVLjZMebfVDRnbHWNvw@mail.gmail.com>
To: Brian E Carpenter <brian.e.carpenter@gmail.com>
Content-Type: multipart/alternative; boundary=001a1147bea0c5d3950535a3814d
Archived-At: <https://mailarchive.ietf.org/arch/msg/nmlrg/MD6q4DHG519TpyxSF_IcFbKtYZw>
Cc: Michele Zorzi <zorzi@ing.unife.it>, nmlrg@irtf.org
Subject: Re: [Nmlrg] recent paper on cognition-based networks
X-BeenThere: nmlrg@irtf.org
X-Mailman-Version: 2.1.17
Precedence: list
List-Id: Network Machine Learning Research Group <nmlrg.irtf.org>
List-Unsubscribe: <https://www.irtf.org/mailman/options/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=unsubscribe>
List-Archive: <https://mailarchive.ietf.org/arch/browse/nmlrg/>
List-Post: <mailto:nmlrg@irtf.org>
List-Help: <mailto:nmlrg-request@irtf.org?subject=help>
List-Subscribe: <https://www.irtf.org/mailman/listinfo/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=subscribe>
X-List-Received-Date: Sun, 19 Jun 2016 15:47:40 -0000

--001a1147bea0c5d3950535a3814d
Content-Type: text/plain; charset=UTF-8

Dear all

with respect to the article of Michele and to the comments of Brian, I have
a remark, primarily, on the relation of intent and ML techniques (hope I
got the question correctly).

I will start from the second, that of the relation of intent and
unsupervized machine learning techniques.

In general, my view regarding the relation of intent and ML is that it
depends on the scenarios, on what we are trying to solve. I will take the
example as a basis.

-- The declaration of intent may designate "clusters", and, therefore,
imply that supervized learning techniques are enough. E.g., situations
addressed and network configurations used can be split into clusters that
yield jitter above or below the indicated threshold.

-- The declaration of intent may also leave degrees of freedom. In this
case unsupervized learning may be most appropriate. E.g., we may be
interested in aspects further to the jitter entailed by a situation
addressed and the configuration used. Such aspects may be the energy
consumption, the spectrum needed, etc.

Further views are welcome, and of course further elaboration is possible.

Best regards
Panagiotis


On Tue, Jun 14, 2016 at 5:01 AM, Brian E Carpenter <
brian.e.carpenter@gmail.com> wrote:

> Michele,
>
> I read the paper with interest; it even cites an article by one of
> my colleagues in Auckland (Aniket Mahani).
>
> I have a practical question. You correctly observe that computing
> power is now cheap enough to consider such techniques, but I'm
> wondering what that really means. Can we imagine ML inside
> high-end routers? Line speed firewalls? Or will it be confined
> to specialised nodes that serve up ML results to network devices?
>
> Another question is more abstract. You focus on unsupervised
> learning where (if I understand correctly) the features that are
> learned have no direct relationship to features that humans
> consider important. In that situation, how can we translate what
> we are calling "Intent" in the IETF work on autonomics into
> goals for the ML-based system? Or to give a concrete example,
> suppose there is an Intent statement that jitter (more precisely,
> Packet Delay Variation, RFC3393) for voice streams must be less
> than N milliseconds. How do we relate that to ML-derived features
> in a packet stream?
>
> Regards
>    Brian Carpenter
>
> On 24/05/2016 23:52, Michele Zorzi wrote:
> >
> > Hi all,
> >
> > I would like to point you to a recent paper we published, which tries to
> scope out some vision for cognitive networking beyond
> > the more traditional view of cognitive radio. Please find it on
> IEEExplore (it is open access) at
> > http://ieeexplore.ieee.org/xpl/abstractCitations.jsp?arnumber=7217798
> >
> > I will very much welcome your comments and feeback, especially about the
> practical feasibility of this approach. I will also
> > welcome expressions of interest to collaborate on furthering this into
> something more concrete.
> >
> > Thanks a lot,
> >
> > Michele
> >
> > _______________________________________________
> > nmlrg mailing list
> > nmlrg@irtf.org
> > https://www.irtf.org/mailman/listinfo/nmlrg
> >
>
> _______________________________________________
> nmlrg mailing list
> nmlrg@irtf.org
> https://www.irtf.org/mailman/listinfo/nmlrg
>



-- 
Panagiotis Demestichas
Professor
University of Piraeus
School of Information and Communication Technology
Department of Digital Systems
Telecommunication Networks and integrated Services (TNS) Laboratory
Office:         +302104142758
Laboratory: +302104142749
Cellular:      +306944243856
Skype:        pdemest
Office: 150 Androutsou, 18532, Piraeus, GREECE
Mail :  80 Karaoli and Dimitriou, 18534, Piraeus, GREECE

--001a1147bea0c5d3950535a3814d
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr">Dear all<div><br></div><div>with respect to the article of=
 Michele and to the comments of Brian, I have a remark, primarily, on the r=
elation of intent and ML techniques (hope I got the question correctly).</d=
iv><div><br></div><div>I will start from the second, that of the relation o=
f intent and unsupervized machine learning techniques.=C2=A0</div><div><br>=
</div><div>In general, my view regarding the relation of intent and ML is t=
hat it depends on the scenarios, on what we are trying to solve. I will tak=
e the example as a basis.=C2=A0</div><div><br></div><div>--=C2=A0The declar=
ation of intent=C2=A0may designate &quot;clusters&quot;, and, therefore, im=
ply that supervized learning techniques are enough. E.g., situations addres=
sed and network configurations used can be split into clusters that yield j=
itter above or below the indicated threshold.</div><div><br></div><div>-- T=
he declaration of intent may also leave degrees of freedom. In this case un=
supervized learning may be most appropriate. E.g., we may be interested in =
aspects further to the jitter entailed by a situation addressed and the con=
figuration used. Such aspects may be the energy consumption, the spectrum n=
eeded, etc.</div><div><br></div><div>Further views are welcome, and of cour=
se further elaboration is possible.</div><div><br></div><div>Best regards</=
div><div>Panagiotis</div><div><br></div><div class=3D"gmail_extra"><br><div=
 class=3D"gmail_quote">On Tue, Jun 14, 2016 at 5:01 AM, Brian E Carpenter <=
span dir=3D"ltr">&lt;<a href=3D"mailto:brian.e.carpenter@gmail.com" target=
=3D"_blank">brian.e.carpenter@gmail.com</a>&gt;</span> wrote:<br><blockquot=
e class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1px #ccc sol=
id;padding-left:1ex">Michele,<br>
<br>
I read the paper with interest; it even cites an article by one of<br>
my colleagues in Auckland (Aniket Mahani).<br>
<br>
I have a practical question. You correctly observe that computing<br>
power is now cheap enough to consider such techniques, but I&#39;m<br>
wondering what that really means. Can we imagine ML inside<br>
high-end routers? Line speed firewalls? Or will it be confined<br>
to specialised nodes that serve up ML results to network devices?<br>
<br>
Another question is more abstract. You focus on unsupervised<br>
learning where (if I understand correctly) the features that are<br>
learned have no direct relationship to features that humans<br>
consider important. In that situation, how can we translate what<br>
we are calling &quot;Intent&quot; in the IETF work on autonomics into<br>
goals for the ML-based system? Or to give a concrete example,<br>
suppose there is an Intent statement that jitter (more precisely,<br>
Packet Delay Variation, RFC3393) for voice streams must be less<br>
than N milliseconds. How do we relate that to ML-derived features<br>
in a packet stream?<br>
<br>
Regards<br>
=C2=A0 =C2=A0Brian Carpenter<br>
<br>
On 24/05/2016 23:52, Michele Zorzi wrote:<br>
&gt;<br>
&gt; Hi all,<br>
&gt;<br>
&gt; I would like to point you to a recent paper we published, which tries =
to scope out some vision for cognitive networking beyond<br>
&gt; the more traditional view of cognitive radio. Please find it on IEEExp=
lore (it is open access) at<br>
&gt; <a href=3D"http://ieeexplore.ieee.org/xpl/abstractCitations.jsp?arnumb=
er=3D7217798" rel=3D"noreferrer" target=3D"_blank">http://ieeexplore.ieee.o=
rg/xpl/abstractCitations.jsp?arnumber=3D7217798</a><br>
&gt;<br>
&gt; I will very much welcome your comments and feeback, especially about t=
he practical feasibility of this approach. I will also<br>
&gt; welcome expressions of interest to collaborate on furthering this into=
 something more concrete.<br>
&gt;<br>
&gt; Thanks a lot,<br>
&gt;<br>
&gt; Michele<br>
&gt;<br>
&gt; _______________________________________________<br>
&gt; nmlrg mailing list<br>
&gt; <a href=3D"mailto:nmlrg@irtf.org" target=3D"_blank">nmlrg@irtf.org</a>=
<br>
&gt; <a href=3D"https://www.irtf.org/mailman/listinfo/nmlrg" rel=3D"norefer=
rer" target=3D"_blank">https://www.irtf.org/mailman/listinfo/nmlrg</a><br>
&gt;<br>
<br>
_______________________________________________<br>
nmlrg mailing list<br>
<a href=3D"mailto:nmlrg@irtf.org" target=3D"_blank">nmlrg@irtf.org</a><br>
<a href=3D"https://www.irtf.org/mailman/listinfo/nmlrg" rel=3D"noreferrer" =
target=3D"_blank">https://www.irtf.org/mailman/listinfo/nmlrg</a><br>
</blockquote></div><br></div><br clear=3D"all"><div><br></div>-- <br><div c=
lass=3D"gmail_signature" data-smartmail=3D"gmail_signature"><div dir=3D"ltr=
"><div><div dir=3D"ltr"><div><div dir=3D"ltr">Panagiotis Demestichas<br>Pro=
fessor<br>University of Piraeus<br>School of Information and Communication =
Technology<br>Department of Digital Systems<br>Telecommunication Networks a=
nd integrated Services (TNS) Laboratory<br>Office: =C2=A0 =C2=A0 =C2=A0 =C2=
=A0 +302104142758<br>Laboratory: +302104142749<br>Cellular:=C2=A0=C2=A0=C2=
=A0=C2=A0=C2=A0 +306944243856</div><div dir=3D"ltr">Skype: =C2=A0 =C2=A0 =
=C2=A0 =C2=A0pdemest<br>Office: 150 Androutsou, 18532, Piraeus, GREECE<br>M=
ail :=C2=A0 80 Karaoli and Dimitriou, 18534, Piraeus, GREECE<br></div></div=
></div></div></div></div></div>

--001a1147bea0c5d3950535a3814d--


From nobody Fri Jun 24 09:10:29 2016
Return-Path: <agenda@ietf.org>
X-Original-To: nmlrg@irtf.org
Delivered-To: nmlrg@ietfa.amsl.com
Received: from ietfa.amsl.com (localhost [IPv6:::1]) by ietfa.amsl.com (Postfix) with ESMTP id DD73212DD71; Fri, 24 Jun 2016 09:01:02 -0700 (PDT)
MIME-Version: 1.0
Content-Type: text/plain; charset="utf-8"
Content-Transfer-Encoding: 7bit
From: "\"IETF Secretariat\"" <agenda@ietf.org>
To: <nmlrg-chairs@ietf.org>, <jiangsheng@huawei.com>
X-Test-IDTracker: no
X-IETF-IDTracker: 6.24.0
Auto-Submitted: auto-generated
Precedence: bulk
Message-ID: <20160624160102.10933.46959.idtracker@ietfa.amsl.com>
Date: Fri, 24 Jun 2016 09:01:02 -0700
Archived-At: <https://mailarchive.ietf.org/arch/msg/nmlrg/ZlHEwb-1sdB1HVDXMKI7s2UKMHg>
Cc: nmlrg@irtf.org, irtf-chair@irtf.org
Subject: [Nmlrg] nmlrg - Requested session has been scheduled for IETF 96
X-BeenThere: nmlrg@irtf.org
X-Mailman-Version: 2.1.17
List-Id: Network Machine Learning Research Group <nmlrg.irtf.org>
List-Unsubscribe: <https://www.irtf.org/mailman/options/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=unsubscribe>
List-Archive: <https://mailarchive.ietf.org/arch/browse/nmlrg/>
List-Post: <mailto:nmlrg@irtf.org>
List-Help: <mailto:nmlrg-request@irtf.org?subject=help>
List-Subscribe: <https://www.irtf.org/mailman/listinfo/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=subscribe>
X-List-Received-Date: Fri, 24 Jun 2016 16:01:03 -0000

Dear Sheng Jiang,

The session(s) that you have requested have been scheduled.
Below is the scheduled session information followed by
the original request. 

nmlrg Session 1 (2:00:00)
    Monday, Afternoon Session II 1540-1740
    Room Name: Potsdam II size: 175
    ---------------------------------------------
    


Request Information:


---------------------------------------------------------
Working Group Name: Proposed Network Machine Learning Research Group
Area Name: IRTF
Session Requester: Sheng Jiang

Number of Sessions: 1
Length of Session(s):  2 Hours
Number of Attendees: 100
Conflicts to Avoid: 
 First Priority: anima dhc sdnrg nfvrg nmrg
 Second Priority: 6man v6ops supa intarea
 Third Priority: netconf softwire


Special Requests:
  
---------------------------------------------------------


From nobody Tue Jun 28 06:04:10 2016
Return-Path: <jiangsheng@huawei.com>
X-Original-To: nmlrg@ietfa.amsl.com
Delivered-To: nmlrg@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 4195E12DEA1 for <nmlrg@ietfa.amsl.com>; Tue, 28 Jun 2016 06:04:02 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -5.647
X-Spam-Level: 
X-Spam-Status: No, score=-5.647 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, RCVD_IN_DNSWL_MED=-2.3, RCVD_IN_MSPIKE_H4=-0.01, RCVD_IN_MSPIKE_WL=-0.01, RP_MATCHES_RCVD=-1.426, SPF_PASS=-0.001] autolearn=ham autolearn_force=no
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id P5MHF59Zj6zN for <nmlrg@ietfa.amsl.com>; Tue, 28 Jun 2016 06:03:59 -0700 (PDT)
Received: from szxga02-in.huawei.com (szxga02-in.huawei.com [119.145.14.65]) (using TLSv1 with cipher RC4-SHA (128/128 bits)) (No client certificate requested) by ietfa.amsl.com (Postfix) with ESMTPS id B312312D767 for <nmlrg@irtf.org>; Tue, 28 Jun 2016 06:03:58 -0700 (PDT)
Received: from 172.24.1.136 (EHLO nkgeml411-hub.china.huawei.com) ([172.24.1.136]) by szxrg02-dlp.huawei.com (MOS 4.3.7-GA FastPath queued) with ESMTP id DJM44865; Tue, 28 Jun 2016 21:03:50 +0800 (CST)
Received: from NKGEML515-MBX.china.huawei.com ([fe80::a54a:89d2:c471:ff]) by nkgeml411-hub.china.huawei.com ([10.98.56.70]) with mapi id 14.03.0235.001; Tue, 28 Jun 2016 21:03:48 +0800
From: Sheng Jiang <jiangsheng@huawei.com>
To: "nmlrg@irtf.org" <nmlrg@irtf.org>
Thread-Topic: slides of NMLRG #3, June 27th, Athens
Thread-Index: AdHRPYJNCddnKYQuSGiiJXu27XRyBw==
Date: Tue, 28 Jun 2016 13:03:47 +0000
Message-ID: <5D36713D8A4E7348A7E10DF7437A4B927CA8939C@NKGEML515-MBX.china.huawei.com>
Accept-Language: en-GB, zh-CN, en-US
Content-Language: en-GB
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
x-originating-ip: [10.47.70.225]
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: quoted-printable
MIME-Version: 1.0
X-CFilter-Loop: Reflected
X-Mirapoint-Virus-RAPID-Raw: score=unknown(0), refid=str=0001.0A090202.577275B6.0147, ss=1, re=0.000, recu=0.000, reip=0.000,  cl=1, cld=1, fgs=0, ip=0.0.0.0, so=2013-06-18 04:22:30, dmn=2013-03-21 17:37:32
X-Mirapoint-Loop-Id: 3a1c1e93f8815b535d0dffd37c93d42e
Archived-At: <https://mailarchive.ietf.org/arch/msg/nmlrg/KrSknIAtO_upyL-2TDxgjMG2N58>
Subject: [Nmlrg] slides of NMLRG #3, June 27th, Athens
X-BeenThere: nmlrg@irtf.org
X-Mailman-Version: 2.1.17
Precedence: list
List-Id: Network Machine Learning Research Group <nmlrg.irtf.org>
List-Unsubscribe: <https://www.irtf.org/mailman/options/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=unsubscribe>
List-Archive: <https://mailarchive.ietf.org/arch/browse/nmlrg/>
List-Post: <mailto:nmlrg@irtf.org>
List-Help: <mailto:nmlrg-request@irtf.org?subject=help>
List-Subscribe: <https://www.irtf.org/mailman/listinfo/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=subscribe>
X-List-Received-Date: Tue, 28 Jun 2016 13:04:02 -0000

Hi, nmlrg,

All slides that have been presented in our NMLRG #3 meeting, June 27th, 201=
6, Athens, Greece, with EUCNC2016, have been uploaded. They can be accessed=
 through below link

https://www.ietf.org/proceedings/interim/2016/06/27/nmlrg/proceedings.html

Best regards,

Sheng=


From nobody Tue Jun 28 07:03:58 2016
Return-Path: <jiangsheng@huawei.com>
X-Original-To: nmlrg@ietfa.amsl.com
Delivered-To: nmlrg@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 03CE012D1A4 for <nmlrg@ietfa.amsl.com>; Tue, 28 Jun 2016 07:03:57 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -5.647
X-Spam-Level: 
X-Spam-Status: No, score=-5.647 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, RCVD_IN_DNSWL_MED=-2.3, RCVD_IN_MSPIKE_H3=-0.01, RCVD_IN_MSPIKE_WL=-0.01, RP_MATCHES_RCVD=-1.426, SPF_PASS=-0.001] autolearn=ham autolearn_force=no
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 0Edcd2oEwUWI for <nmlrg@ietfa.amsl.com>; Tue, 28 Jun 2016 07:03:53 -0700 (PDT)
Received: from szxga01-in.huawei.com (szxga01-in.huawei.com [58.251.152.64]) (using TLSv1 with cipher RC4-SHA (128/128 bits)) (No client certificate requested) by ietfa.amsl.com (Postfix) with ESMTPS id 60C1B12D198 for <nmlrg@irtf.org>; Tue, 28 Jun 2016 07:03:42 -0700 (PDT)
Received: from 172.24.1.47 (EHLO nkgeml412-hub.china.huawei.com) ([172.24.1.47]) by szxrg01-dlp.huawei.com (MOS 4.3.7-GA FastPath queued) with ESMTP id DMX50391; Tue, 28 Jun 2016 22:03:37 +0800 (CST)
Received: from NKGEML515-MBX.china.huawei.com ([fe80::a54a:89d2:c471:ff]) by nkgeml412-hub.china.huawei.com ([10.98.56.73]) with mapi id 14.03.0235.001; Tue, 28 Jun 2016 22:03:30 +0800
From: Sheng Jiang <jiangsheng@huawei.com>
To: "nmlrg@irtf.org" <nmlrg@irtf.org>
Thread-Topic: links to each slide of presentations//RE: slides of NMLRG #3, June 27th, Athens
Thread-Index: AQHR0UXZfny1k5v9oUOv37+HaylYtw==
Date: Tue, 28 Jun 2016 14:03:29 +0000
Message-ID: <5D36713D8A4E7348A7E10DF7437A4B927CA893E9@NKGEML515-MBX.china.huawei.com>
Accept-Language: en-GB, zh-CN, en-US
Content-Language: en-GB
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
x-originating-ip: [10.47.70.225]
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: quoted-printable
MIME-Version: 1.0
X-CFilter-Loop: Reflected
X-Mirapoint-Virus-RAPID-Raw: score=unknown(0), refid=str=0001.0A020203.577283BA.0184, ss=1, re=0.000, recu=0.000, reip=0.000,  cl=1, cld=1, fgs=0, ip=0.0.0.0, so=2013-06-18 04:22:30, dmn=2013-03-21 17:37:32
X-Mirapoint-Loop-Id: 4e0e67f197324c0b59ed16082321e5c7
Archived-At: <https://mailarchive.ietf.org/arch/msg/nmlrg/wu-F-MGmLdHvZy5RTAmWTfRJawk>
Subject: [Nmlrg] links to each slide of presentations//RE: slides of NMLRG #3, June 27th, Athens
X-BeenThere: nmlrg@irtf.org
X-Mailman-Version: 2.1.17
Precedence: list
List-Id: Network Machine Learning Research Group <nmlrg.irtf.org>
List-Unsubscribe: <https://www.irtf.org/mailman/options/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=unsubscribe>
List-Archive: <https://mailarchive.ietf.org/arch/browse/nmlrg/>
List-Post: <mailto:nmlrg@irtf.org>
List-Help: <mailto:nmlrg-request@irtf.org?subject=help>
List-Subscribe: <https://www.irtf.org/mailman/listinfo/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=subscribe>
X-List-Received-Date: Tue, 28 Jun 2016 14:03:57 -0000

Oops... The proceeding page for interim meeting seems not be as intelligent=
 as the proceeding pages of IETF meetings. Our proceeding page does not aut=
onomically show the slides. I have sent an email to IETF secretary to ask t=
hem to fix it. Meanwhile, in this email, here is the links for each present=
ations:

Slide Filename Edit Replace Delete=20
Chair Slides=20
https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interi=
m-2016-nmlrg-01-0.pdf

Introduction to Network Machine Learning & NMLRG
https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interi=
m-2016-nmlrg-01-1.pdf

Data Collection and Analysis At High Security Lab
https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interi=
m-2016-nmlrg-01-2.pdf

Use Cases of Applying Machine Learning Mechanism with Network Traffic=20
https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interi=
m-2016-nmlrg-01-3.pdf

Mobile network state characterization and prediction
https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interi=
m-2016-nmlrg-01-4.pdf

Learning how to route=20
https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interi=
m-2016-nmlrg-01-5.pdf

Regards,

Sheng
________________________________________
From: nmlrg [nmlrg-bounces@irtf.org] on behalf of Sheng Jiang [jiangsheng@h=
uawei.com]
Sent: 28 June 2016 21:03
To: nmlrg@irtf.org
Subject: [Nmlrg] slides of NMLRG #3, June 27th, Athens

Hi, nmlrg,

All slides that have been presented in our NMLRG #3 meeting, June 27th, 201=
6, Athens, Greece, with EUCNC2016, have been uploaded. They can be accessed=
 through below link

https://www.ietf.org/proceedings/interim/2016/06/27/nmlrg/proceedings.html

Best regards,

Sheng
_______________________________________________
nmlrg mailing list
nmlrg@irtf.org
https://www.irtf.org/mailman/listinfo/nmlrg=


From nobody Tue Jun 28 08:30:46 2016
Return-Path: <dmm@1-4-5.net>
X-Original-To: nmlrg@ietfa.amsl.com
Delivered-To: nmlrg@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id ECE9012D52F for <nmlrg@ietfa.amsl.com>; Tue, 28 Jun 2016 08:30:44 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.599
X-Spam-Level: 
X-Spam-Status: No, score=-2.599 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, HTML_MESSAGE=0.001, RCVD_IN_DNSWL_LOW=-0.7] autolearn=ham autolearn_force=no
Authentication-Results: ietfa.amsl.com (amavisd-new); dkim=pass (2048-bit key) header.d=1-4-5-net.20150623.gappssmtp.com
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id uUUNmPafL7Ri for <nmlrg@ietfa.amsl.com>; Tue, 28 Jun 2016 08:30:40 -0700 (PDT)
Received: from mail-it0-x22c.google.com (mail-it0-x22c.google.com [IPv6:2607:f8b0:4001:c0b::22c]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by ietfa.amsl.com (Postfix) with ESMTPS id 9C8C512D541 for <nmlrg@irtf.org>; Tue, 28 Jun 2016 08:30:40 -0700 (PDT)
Received: by mail-it0-x22c.google.com with SMTP id f6so18031594ith.0 for <nmlrg@irtf.org>; Tue, 28 Jun 2016 08:30:40 -0700 (PDT)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1-4-5-net.20150623.gappssmtp.com; s=20150623; h=mime-version:in-reply-to:references:from:date:message-id:subject:to :cc; bh=DdGGVOsQcvlGY8mhpIoNF0wkE6r9DWdDXMaTCBoIfSY=; b=FrgTB++f+qFynf7LXKxdCvJY/82y4SHGVPHumAezPPEKK0TIf8yQ28Om5pJsNrOGx0 u/KsU3TkChz4LCnuI9xL1RBsY8BDK1+ETJBHVfkrQcXZYNWnv1CBnQujZYiIKodhewCw ODQGXR0LQpNzKAU1RQawDc1QDW9rTs8412QJ8HwKFFP6XpavPYRGc602ZLG9TJOqn5AT MrrfjZwgST4B9UUHWCFZoS1RT57/WA3pbIXvw9Nv79R/aanVgvzKofq+6ztKb5c6E/nv QBSQTA175bY8XIn81L/9V+n8rThtThyE+T9966iMWMKYpRt9LroegGcuxLBONsMJ7eu0 4ncA==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:from:date :message-id:subject:to:cc; bh=DdGGVOsQcvlGY8mhpIoNF0wkE6r9DWdDXMaTCBoIfSY=; b=MWCHE6UPFj98ObPFQ64qA160LEXjt9lAqMNKkp8B56iKwith2uHMBbiXrk53EZZVmE 8Ux046fk6nRtBs2DUsLFdfKgujZXztFjtdFnnazAlyHAr3N0Bs03pat1zxlIueJbWmGZ yM5gF+aejjmcC/HFAI96IoZZJrW/0C3V2b3Efu3k4RpAdLY+Ch2MRrHLTaLjvsKsq6A7 HGlW3ptqVXmQODr2M3rz1rr9KXhzdFtfDxuhfqSjtxxBUBk6yO0hllpjcXTjWwWfSe7Q hi+Gav1MC8MGWvBgDHu2L8DZuO2VBzvmaujfv2dfCkrO8vPdtijS0MRR98y0VWHAfP+q AB7g==
X-Gm-Message-State: ALyK8tJox6lREEddsbaLjTiKf1nJgPI146QiUtGgvwuKLZGpCS4N8f69vCpLFrjEqhHJdXzSUK8v8Uv0Qu31VQ==
X-Received: by 10.36.61.204 with SMTP id n195mr15862115itn.92.1467127839758; Tue, 28 Jun 2016 08:30:39 -0700 (PDT)
MIME-Version: 1.0
Received: by 10.64.148.133 with HTTP; Tue, 28 Jun 2016 08:30:38 -0700 (PDT)
X-Originating-IP: [128.223.156.253]
In-Reply-To: <5D36713D8A4E7348A7E10DF7437A4B927CA893E9@NKGEML515-MBX.china.huawei.com>
References: <5D36713D8A4E7348A7E10DF7437A4B927CA893E9@NKGEML515-MBX.china.huawei.com>
From: David Meyer <dmm@1-4-5.net>
Date: Tue, 28 Jun 2016 08:30:38 -0700
Message-ID: <CAHiKxWhfdukjRVnSKnbLwMwamiqZYoCD350BoNqO4njqkQoU1g@mail.gmail.com>
To: Sheng Jiang <jiangsheng@huawei.com>
Content-Type: multipart/alternative; boundary=001a1144404ebacac80536585173
Archived-At: <https://mailarchive.ietf.org/arch/msg/nmlrg/eG98vUN3EaAiyeVGIkcv1r9xDYw>
Cc: "nmlrg@irtf.org" <nmlrg@irtf.org>
Subject: Re: [Nmlrg] links to each slide of presentations//RE: slides of NMLRG #3, June 27th, Athens
X-BeenThere: nmlrg@irtf.org
X-Mailman-Version: 2.1.17
Precedence: list
List-Id: Network Machine Learning Research Group <nmlrg.irtf.org>
List-Unsubscribe: <https://www.irtf.org/mailman/options/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=unsubscribe>
List-Archive: <https://mailarchive.ietf.org/arch/browse/nmlrg/>
List-Post: <mailto:nmlrg@irtf.org>
List-Help: <mailto:nmlrg-request@irtf.org?subject=help>
List-Subscribe: <https://www.irtf.org/mailman/listinfo/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=subscribe>
X-List-Received-Date: Tue, 28 Jun 2016 15:30:45 -0000

--001a1144404ebacac80536585173
Content-Type: text/plain; charset=UTF-8

Sheng,

Thanks for the pointers. One thing I notice is that we don't have much on
what the characteristics of network data might be, and as such, what kinds
of existing learning algorithms might be suitable and where additional
development might be required. For example, is the collected data IID? Is
it time series? Is the underlying distribution stationary? If not, how does
that constrain algorithms we might use or develop? For example, how does a
given algorithm deal with concept drift/internal covariate shift (in the
case of DNNs; see e.g., batch normalization [0]). There are many other such
questions, such as is the data categorical (e.g., ports, IP addresses) or
is it continuous/discrete (e.g., counters). And if the data in question is
categorical, what is the cardinality of the categories (this will inform
how such data can be encoded); in the case of IP addresses we can't really
one-hot encode addresses because their cardinality is too large (2^32 or
2^128); this has implications for how we build classifiers (in particular,
for softmax layers in DNNs of various kinds).

Related to the above is the question of features? What are good features
for networking? Where do they come from? Are they domain specific? Can we
learn features the way a DNNs does in the network space? Can we use
autoencoders to discover such features? Or can we use GANs to train DNNs
for network classification tasks in an unsupervised manner? Are there
other, non-ad-hoc (well founded) methods we can use, or is every use case a
one-off (one would hope not).

We can carry the same kinds of analyzes to the algorithms applied. For
example, while something k-means is an effective way to get a feeling for
how continuous/discrete hangs together, if our data is categorical
statistical clustering approaches such as LDA might provide a more
well-founded approach (of course, as with most Bayesian techniques is the
question of approximate inference arises since in most interesting cases
the integral that we need to solve, namely the marginal probability data
isn't tractable so we need to resort to MCMC or more likely variational
inference). And what about the use of SGD/batch normalization etc with
DNNs, and perhaps more importantly, can we use network data to train DNN
policy networks for reinforcement learning like we saw in deep Q-learning
and AlphaGo?

These comments are all by way of saying that we don't have a solid
theoretical understanding (yet) of how techniques that have been so
successful in other domains (e.g., DNNs for perceptual tasks) generalize to
networking use cases. We will need this understanding if our goal is to
provide accurate, repeatable, and explainable results.

In order to accomplish all of this we need, as I have been saying , not
only a good understanding of how these algorithms work but also
standardized data sets and associated benchmarks so we can tell if we are
making progress (or even if our techniques work). Analogies here include
MNIST and ImageNet and their associated benchmarks, among others. As
mentioned  standardized data sets are key to making progress in the ML for
networking space (otherwise how do you know your technique works and/or
improves on another techniques?). One might assume that these data sets
would need to be labeled (as supervised learning is where most of the
progress is being made these days), but not necessarily; Generative
Adversarial Networks (GANs) have emerged as a new way to train DNNs in an
unsupervised manner (this is moving very rapidly; see e.g.,
https://openai.com/blog/generative-models/).

The summary here is that the "distance" between theory and practice in ML
is effectively zero right now due to the incredible rate of progress in the
field; this means we need  to understand both sides of the theory/practice
coin in order to be effective. None of the slide decks provide much
background on what the proposed algorithms are, how they work, or why they
should be expected to work on network data.

Finally, if you are interested in LDA or other algorithms there are a few
short explanatory pieces I have written for my team on
http://www.1-4-5.net/~dmm/ml (works in progress).

Thanks,

Dave

[0] https://arxiv.org/pdf/1502.03167.pdf

On Tue, Jun 28, 2016 at 7:03 AM, Sheng Jiang <jiangsheng@huawei.com> wrote:

> Oops... The proceeding page for interim meeting seems not be as
> intelligent as the proceeding pages of IETF meetings. Our proceeding page
> does not autonomically show the slides. I have sent an email to IETF
> secretary to ask them to fix it. Meanwhile, in this email, here is the
> links for each presentations:
>
> Slide Filename Edit Replace Delete
> Chair Slides
>
> https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-0.pdf
>
> Introduction to Network Machine Learning & NMLRG
>
> https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-1.pdf
>
> Data Collection and Analysis At High Security Lab
>
> https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-2.pdf
>
> Use Cases of Applying Machine Learning Mechanism with Network Traffic
>
> https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-3.pdf
>
> Mobile network state characterization and prediction
>
> https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-4.pdf
>
> Learning how to route
>
> https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-5.pdf
>
> Regards,
>
> Sheng
> ________________________________________
> From: nmlrg [nmlrg-bounces@irtf.org] on behalf of Sheng Jiang [
> jiangsheng@huawei.com]
> Sent: 28 June 2016 21:03
> To: nmlrg@irtf.org
> Subject: [Nmlrg] slides of NMLRG #3, June 27th, Athens
>
> Hi, nmlrg,
>
> All slides that have been presented in our NMLRG #3 meeting, June 27th,
> 2016, Athens, Greece, with EUCNC2016, have been uploaded. They can be
> accessed through below link
>
> https://www.ietf.org/proceedings/interim/2016/06/27/nmlrg/proceedings.html
>
> Best regards,
>
> Sheng
> _______________________________________________
> nmlrg mailing list
> nmlrg@irtf.org
> https://www.irtf.org/mailman/listinfo/nmlrg
> _______________________________________________
> nmlrg mailing list
> nmlrg@irtf.org
> https://www.irtf.org/mailman/listinfo/nmlrg
>

--001a1144404ebacac80536585173
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr">Sheng,<div><br></div><div>Thanks for the pointers. One thi=
ng I notice is that we don&#39;t have much on what the characteristics of n=
etwork data might be, and as such, what kinds of existing learning algorith=
ms might be suitable and where additional development might be required. Fo=
r example, is the collected data IID? Is it time series? Is the underlying =
distribution stationary? If not, how does that constrain algorithms we migh=
t use or develop? For example, how does a given algorithm deal with concept=
 drift/internal covariate shift (in the case of DNNs; see e.g., batch norma=
lization [0]). There are many other such questions, such as is the data cat=
egorical (e.g., ports, IP addresses) or is it=C2=A0continuous/discrete (e.g=
., counters). And if the data in question is categorical, what is the=C2=A0=
cardinality of the categories (this will inform how such data can be encode=
d); in the case of IP addresses we can&#39;t really one-hot encode addresse=
s because their cardinality is too large (2^32 or 2^128); this has implicat=
ions for how we build classifiers (in particular, for softmax layers in DNN=
s of various kinds).</div><div><br></div><div>Related to the above is the q=
uestion of features? What are good features for networking? Where do they c=
ome from? Are they domain specific? Can we learn features the way a DNNs do=
es in the network space? Can we use autoencoders to discover such features?=
 Or can we use GANs to train DNNs for network classification tasks in an un=
supervised manner? Are there other, non-ad-hoc (well founded) methods we ca=
n use, or is every use case a one-off (one would hope not).</div><div><br><=
/div><div>We can carry the same kinds of analyzes to the algorithms applied=
. For example, while something k-means is an effective way to get a feeling=
 for how=C2=A0continuous/discrete hangs together, if our data is categorica=
l statistical clustering approaches such as LDA might provide a more well-f=
ounded approach (of course, as with most Bayesian techniques is the questio=
n of approximate inference arises since in most interesting cases the integ=
ral that we need to solve, namely the marginal probability data isn&#39;t t=
ractable so we need to resort to MCMC or more likely=C2=A0variational infer=
ence). And what about the use of SGD/batch normalization etc with DNNs, and=
 perhaps more importantly, can we use network data to train DNN policy netw=
orks for reinforcement learning like we saw in deep Q-learning and AlphaGo?=
</div><div><br></div><div>These comments are all by way of saying that we d=
on&#39;t have a solid theoretical understanding (yet) of how techniques tha=
t have been so successful in other domains (e.g., DNNs for perceptual tasks=
) generalize to networking use cases. We will need this understanding if ou=
r goal is to provide accurate, repeatable, and explainable results.=C2=A0</=
div><div><br></div><div>In order to accomplish all of this we need, as I ha=
ve been saying , not only a good understanding of how these algorithms work=
 but also standardized data sets and associated benchmarks so we can tell i=
f we are making progress (or even if our techniques work). Analogies here i=
nclude MNIST and ImageNet and their associated benchmarks, among others. As=
 mentioned =C2=A0standardized data sets are key to making progress in the M=
L for networking space (otherwise how do you know your technique works and/=
or improves on another techniques?). One might assume that these data sets =
would need to be labeled (as supervised learning is where most of the progr=
ess is being made these days), but not necessarily; Generative Adversarial =
Networks (GANs) have emerged as a new way to train DNNs in an unsupervised =
manner (this is moving very rapidly; see e.g.,=C2=A0<a href=3D"https://open=
ai.com/blog/generative-models/">https://openai.com/blog/generative-models/<=
/a>).</div><div><br></div><div>The summary here is that the &quot;distance&=
quot; between theory and practice in ML is effectively zero right now due t=
o the incredible rate of progress in the field; this means we need =C2=A0to=
 understand both sides of the theory/practice coin in order to be effective=
. None of the slide decks provide much background on what the proposed algo=
rithms are, how they work, or why they should be expected to work on networ=
k data.</div><div><br></div><div>Finally, if you are interested in LDA or o=
ther algorithms there are a few short explanatory pieces I have written for=
 my team on <a href=3D"http://www.1-4-5.net/~dmm/ml">http://www.1-4-5.net/~=
dmm/ml</a> (works in progress).</div><div><br></div><div>Thanks,</div><div>=
<br>Dave</div><div><br></div><div>[0]=C2=A0<a href=3D"https://arxiv.org/pdf=
/1502.03167.pdf">https://arxiv.org/pdf/1502.03167.pdf</a></div></div><div c=
lass=3D"gmail_extra"><br><div class=3D"gmail_quote">On Tue, Jun 28, 2016 at=
 7:03 AM, Sheng Jiang <span dir=3D"ltr">&lt;<a href=3D"mailto:jiangsheng@hu=
awei.com" target=3D"_blank">jiangsheng@huawei.com</a>&gt;</span> wrote:<br>=
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex">Oops... The proceeding page for interim meet=
ing seems not be as intelligent as the proceeding pages of IETF meetings. O=
ur proceeding page does not autonomically show the slides. I have sent an e=
mail to IETF secretary to ask them to fix it. Meanwhile, in this email, her=
e is the links for each presentations:<br>
<br>
Slide Filename Edit Replace Delete<br>
Chair Slides<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-0.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-0.pdf</a><br>
<br>
Introduction to Network Machine Learning &amp; NMLRG<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-1.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-1.pdf</a><br>
<br>
Data Collection and Analysis At High Security Lab<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-2.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-2.pdf</a><br>
<br>
Use Cases of Applying Machine Learning Mechanism with Network Traffic<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-3.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-3.pdf</a><br>
<br>
Mobile network state characterization and prediction<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-4.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-4.pdf</a><br>
<br>
Learning how to route<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-5.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-5.pdf</a><br>
<br>
Regards,<br>
<br>
Sheng<br>
________________________________________<br>
From: nmlrg [<a href=3D"mailto:nmlrg-bounces@irtf.org">nmlrg-bounces@irtf.o=
rg</a>] on behalf of Sheng Jiang [<a href=3D"mailto:jiangsheng@huawei.com">=
jiangsheng@huawei.com</a>]<br>
Sent: 28 June 2016 21:03<br>
To: <a href=3D"mailto:nmlrg@irtf.org">nmlrg@irtf.org</a><br>
Subject: [Nmlrg] slides of NMLRG #3, June 27th, Athens<br>
<br>
Hi, nmlrg,<br>
<br>
All slides that have been presented in our NMLRG #3 meeting, June 27th, 201=
6, Athens, Greece, with EUCNC2016, have been uploaded. They can be accessed=
 through below link<br>
<br>
<a href=3D"https://www.ietf.org/proceedings/interim/2016/06/27/nmlrg/procee=
dings.html" rel=3D"noreferrer" target=3D"_blank">https://www.ietf.org/proce=
edings/interim/2016/06/27/nmlrg/proceedings.html</a><br>
<br>
Best regards,<br>
<br>
Sheng<br>
_______________________________________________<br>
nmlrg mailing list<br>
<a href=3D"mailto:nmlrg@irtf.org">nmlrg@irtf.org</a><br>
<a href=3D"https://www.irtf.org/mailman/listinfo/nmlrg" rel=3D"noreferrer" =
target=3D"_blank">https://www.irtf.org/mailman/listinfo/nmlrg</a><br>
_______________________________________________<br>
nmlrg mailing list<br>
<a href=3D"mailto:nmlrg@irtf.org">nmlrg@irtf.org</a><br>
<a href=3D"https://www.irtf.org/mailman/listinfo/nmlrg" rel=3D"noreferrer" =
target=3D"_blank">https://www.irtf.org/mailman/listinfo/nmlrg</a><br>
</blockquote></div><br></div>

--001a1144404ebacac80536585173--


From nobody Wed Jun 29 01:10:51 2016
Return-Path: <jiangsheng@huawei.com>
X-Original-To: nmlrg@ietfa.amsl.com
Delivered-To: nmlrg@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id B5BB012D598 for <nmlrg@ietfa.amsl.com>; Wed, 29 Jun 2016 01:10:50 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -5.646
X-Spam-Level: 
X-Spam-Status: No, score=-5.646 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, HTML_MESSAGE=0.001, RCVD_IN_DNSWL_MED=-2.3, RCVD_IN_MSPIKE_H4=-0.01, RCVD_IN_MSPIKE_WL=-0.01, RP_MATCHES_RCVD=-1.426, SPF_PASS=-0.001] autolearn=ham autolearn_force=no
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id ph6_0VFMKDUR for <nmlrg@ietfa.amsl.com>; Wed, 29 Jun 2016 01:10:47 -0700 (PDT)
Received: from szxga02-in.huawei.com (szxga02-in.huawei.com [119.145.14.65]) (using TLSv1 with cipher RC4-SHA (128/128 bits)) (No client certificate requested) by ietfa.amsl.com (Postfix) with ESMTPS id 0E8BA12DA10 for <nmlrg@irtf.org>; Wed, 29 Jun 2016 01:10:45 -0700 (PDT)
Received: from 172.24.1.60 (EHLO nkgeml411-hub.china.huawei.com) ([172.24.1.60]) by szxrg02-dlp.huawei.com (MOS 4.3.7-GA FastPath queued) with ESMTP id DJO16105; Wed, 29 Jun 2016 16:10:33 +0800 (CST)
Received: from NKGEML515-MBX.china.huawei.com ([fe80::a54a:89d2:c471:ff]) by nkgeml411-hub.china.huawei.com ([10.98.56.70]) with mapi id 14.03.0235.001; Wed, 29 Jun 2016 16:10:24 +0800
From: Sheng Jiang <jiangsheng@huawei.com>
To: David Meyer <dmm@1-4-5.net>
Thread-Topic: [Nmlrg] links to each slide of presentations//RE: slides of NMLRG #3, June 27th, Athens
Thread-Index: AQHR0UXZfny1k5v9oUOv37+HaylYt5/+e2IAgAGH4Uw=
Date: Wed, 29 Jun 2016 08:10:24 +0000
Message-ID: <5D36713D8A4E7348A7E10DF7437A4B927CA897D4@NKGEML515-MBX.china.huawei.com>
References: <5D36713D8A4E7348A7E10DF7437A4B927CA893E9@NKGEML515-MBX.china.huawei.com>,  <CAHiKxWhfdukjRVnSKnbLwMwamiqZYoCD350BoNqO4njqkQoU1g@mail.gmail.com>
In-Reply-To: <CAHiKxWhfdukjRVnSKnbLwMwamiqZYoCD350BoNqO4njqkQoU1g@mail.gmail.com>
Accept-Language: en-GB, zh-CN, en-US
Content-Language: en-GB
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
x-originating-ip: [10.47.66.194]
Content-Type: multipart/alternative; boundary="_000_5D36713D8A4E7348A7E10DF7437A4B927CA897D4NKGEML515MBXchi_"
MIME-Version: 1.0
X-CFilter-Loop: Reflected
X-Mirapoint-Virus-RAPID-Raw: score=unknown(0), refid=str=0001.0A020206.5773827D.0013, ss=1, re=0.000, recu=0.000, reip=0.000,  cl=1, cld=1, fgs=0, ip=0.0.0.0, so=2013-06-18 04:22:30, dmn=2013-03-21 17:37:32
X-Mirapoint-Loop-Id: d3da655d6566e22179791efb37db52bc
Archived-At: <https://mailarchive.ietf.org/arch/msg/nmlrg/YTkZ9LcXGma6ZOBh0VoR7UiDQoI>
Cc: "nmlrg@irtf.org" <nmlrg@irtf.org>
Subject: Re: [Nmlrg] links to each slide of presentations//RE: slides of NMLRG #3, June 27th, Athens
X-BeenThere: nmlrg@irtf.org
X-Mailman-Version: 2.1.17
Precedence: list
List-Id: Network Machine Learning Research Group <nmlrg.irtf.org>
List-Unsubscribe: <https://www.irtf.org/mailman/options/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=unsubscribe>
List-Archive: <https://mailarchive.ietf.org/arch/browse/nmlrg/>
List-Post: <mailto:nmlrg@irtf.org>
List-Help: <mailto:nmlrg-request@irtf.org?subject=help>
List-Subscribe: <https://www.irtf.org/mailman/listinfo/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=subscribe>
X-List-Received-Date: Wed, 29 Jun 2016 08:10:50 -0000

--_000_5D36713D8A4E7348A7E10DF7437A4B927CA897D4NKGEML515MBXchi_
Content-Type: text/plain; charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable

Hi, David,



Thanks so much for your email. I am sharing the same concern you have. Actu=
ally, this Monday in our NMLRG #3, Athens, we did discussed this issue rega=
rding to the potential standardized dataset. And we are planning to have an=
 open discussion to regarding to the requirements and validation qualificat=
ion of the potential common dataset in the coming NMLRG #4, Berlin, IETF96.



However, upto now, I have serveral concerns. Firstly, before we could discu=
ss such dataset, we need to narrow our target scenarios to a few very speci=
fic use cases, giving that there are so many various scenarios in the netwo=
rk area and they are so different from each other and require different dat=
aset. Secondly, as we did talk with operators, the real data is the importa=
nt properties for them. So, it would be almost impossible for them to share=
 them. It leaves two possibilities on how to get dataset: a) generating dat=
aset using simulation; b) try to get some partial dataset from operators th=
at have removed their sensenble data. Both way are problemstic, in the risk=
 that a lot of efforts may waste in wrong areas. Thirdly, the availability =
of data may very different in different network environments. It depends on=
 the measurement/congnitive functions that have been deployed or implemente=
d.



Although having the above concerns, I  think we should: a) research on cert=
ain dataset. It would be very useful to prove the machine learning could pe=
rform better or provide more flexibility and adaptibility; b) continue our =
discuss on various network use cases that could benefit from applying machi=
ne learning, in order to serve wider network communities.



Another important point I would like to discuss is that you mentioned "our =
goal is to provide accurate, repeatable, and explainable results." Ideally,=
 we do want this. It is our traditional way for logic-based programming. Ho=
wever, I am not sure that is a doable target with machine learning. Many of=
 machine learning algorithms are based on statistics theory, which may not =
be that much accurate. Some of machine learning algorithms are black box or=
 have very complexicated internal computing process. So, the resulf of thes=
e algorithms may be hard to be explained. Also the process of machine learn=
ing may be hard to be controlled or intervened. Repeatable, yes, we have to=
 have this feature to be widely adopted. However, repeatable may have to co=
me with some level of toleration for the inaccurate.



Best regards,



Sheng



________________________________

From: David Meyer [dmm@1-4-5.net]
Sent: 28 June 2016 23:30
To: Sheng Jiang
Cc: nmlrg@irtf.org
Subject: Re: [Nmlrg] links to each slide of presentations//RE: slides of NM=
LRG #3, June 27th, Athens

Sheng,

Thanks for the pointers. One thing I notice is that we don't have much on w=
hat the characteristics of network data might be, and as such, what kinds o=
f existing learning algorithms might be suitable and where additional devel=
opment might be required. For example, is the collected data IID? Is it tim=
e series? Is the underlying distribution stationary? If not, how does that =
constrain algorithms we might use or develop? For example, how does a given=
 algorithm deal with concept drift/internal covariate shift (in the case of=
 DNNs; see e.g., batch normalization [0]). There are many other such questi=
ons, such as is the data categorical (e.g., ports, IP addresses) or is it c=
ontinuous/discrete (e.g., counters). And if the data in question is categor=
ical, what is the cardinality of the categories (this will inform how such =
data can be encoded); in the case of IP addresses we can't really one-hot e=
ncode addresses because their cardinality is too large (2^32 or 2^128); thi=
s has implications for how we build classifiers (in particular, for softmax=
 layers in DNNs of various kinds).

Related to the above is the question of features? What are good features fo=
r networking? Where do they come from? Are they domain specific? Can we lea=
rn features the way a DNNs does in the network space? Can we use autoencode=
rs to discover such features? Or can we use GANs to train DNNs for network =
classification tasks in an unsupervised manner? Are there other, non-ad-hoc=
 (well founded) methods we can use, or is every use case a one-off (one wou=
ld hope not).

We can carry the same kinds of analyzes to the algorithms applied. For exam=
ple, while something k-means is an effective way to get a feeling for how c=
ontinuous/discrete hangs together, if our data is categorical statistical c=
lustering approaches such as LDA might provide a more well-founded approach=
 (of course, as with most Bayesian techniques is the question of approximat=
e inference arises since in most interesting cases the integral that we nee=
d to solve, namely the marginal probability data isn't tractable so we need=
 to resort to MCMC or more likely variational inference). And what about th=
e use of SGD/batch normalization etc with DNNs, and perhaps more importantl=
y, can we use network data to train DNN policy networks for reinforcement l=
earning like we saw in deep Q-learning and AlphaGo?

These comments are all by way of saying that we don't have a solid theoreti=
cal understanding (yet) of how techniques that have been so successful in o=
ther domains (e.g., DNNs for perceptual tasks) generalize to networking use=
 cases. We will need this understanding if our goal is to provide accurate,=
 repeatable, and explainable results.

In order to accomplish all of this we need, as I have been saying , not onl=
y a good understanding of how these algorithms work but also standardized d=
ata sets and associated benchmarks so we can tell if we are making progress=
 (or even if our techniques work). Analogies here include MNIST and ImageNe=
t and their associated benchmarks, among others. As mentioned  standardized=
 data sets are key to making progress in the ML for networking space (other=
wise how do you know your technique works and/or improves on another techni=
ques?). One might assume that these data sets would need to be labeled (as =
supervised learning is where most of the progress is being made these days)=
, but not necessarily; Generative Adversarial Networks (GANs) have emerged =
as a new way to train DNNs in an unsupervised manner (this is moving very r=
apidly; see e.g., https://openai.com/blog/generative-models/).

The summary here is that the "distance" between theory and practice in ML i=
s effectively zero right now due to the incredible rate of progress in the =
field; this means we need  to understand both sides of the theory/practice =
coin in order to be effective. None of the slide decks provide much backgro=
und on what the proposed algorithms are, how they work, or why they should =
be expected to work on network data.

Finally, if you are interested in LDA or other algorithms there are a few s=
hort explanatory pieces I have written for my team on http://www.1-4-5.net/=
~dmm/ml (works in progress).

Thanks,

Dave

[0] https://arxiv.org/pdf/1502.03167.pdf

On Tue, Jun 28, 2016 at 7:03 AM, Sheng Jiang <jiangsheng@huawei.com<mailto:=
jiangsheng@huawei.com>> wrote:
Oops... The proceeding page for interim meeting seems not be as intelligent=
 as the proceeding pages of IETF meetings. Our proceeding page does not aut=
onomically show the slides. I have sent an email to IETF secretary to ask t=
hem to fix it. Meanwhile, in this email, here is the links for each present=
ations:

Slide Filename Edit Replace Delete
Chair Slides
https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interi=
m-2016-nmlrg-01-0.pdf

Introduction to Network Machine Learning & NMLRG
https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interi=
m-2016-nmlrg-01-1.pdf

Data Collection and Analysis At High Security Lab
https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interi=
m-2016-nmlrg-01-2.pdf

Use Cases of Applying Machine Learning Mechanism with Network Traffic
https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interi=
m-2016-nmlrg-01-3.pdf

Mobile network state characterization and prediction
https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interi=
m-2016-nmlrg-01-4.pdf

Learning how to route
https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interi=
m-2016-nmlrg-01-5.pdf

Regards,

Sheng
________________________________________
From: nmlrg [nmlrg-bounces@irtf.org<mailto:nmlrg-bounces@irtf.org>] on beha=
lf of Sheng Jiang [jiangsheng@huawei.com<mailto:jiangsheng@huawei.com>]
Sent: 28 June 2016 21:03
To: nmlrg@irtf.org<mailto:nmlrg@irtf.org>
Subject: [Nmlrg] slides of NMLRG #3, June 27th, Athens

Hi, nmlrg,

All slides that have been presented in our NMLRG #3 meeting, June 27th, 201=
6, Athens, Greece, with EUCNC2016, have been uploaded. They can be accessed=
 through below link

https://www.ietf.org/proceedings/interim/2016/06/27/nmlrg/proceedings.html

Best regards,

Sheng
_______________________________________________
nmlrg mailing list
nmlrg@irtf.org<mailto:nmlrg@irtf.org>
https://www.irtf.org/mailman/listinfo/nmlrg
_______________________________________________
nmlrg mailing list
nmlrg@irtf.org<mailto:nmlrg@irtf.org>
https://www.irtf.org/mailman/listinfo/nmlrg


--_000_5D36713D8A4E7348A7E10DF7437A4B927CA897D4NKGEML515MBXchi_
Content-Type: text/html; charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable

<html dir=3D"ltr">
<head>
<meta http-equiv=3D"Content-Type" content=3D"text/html; charset=3Diso-8859-=
1">
<style id=3D"owaParaStyle">P {
	MARGIN-TOP: 0px; MARGIN-BOTTOM: 0px
}
</style>
</head>
<body fPStyle=3D"1" ocsi=3D"0">
<div style=3D"direction: ltr;font-family: Tahoma;color: #000000;font-size: =
10pt;">
<p>Hi, David,</p>
<p>&nbsp;</p>
<p>Thanks so much for your email. I am sharing the same concern you have. A=
ctually, this Monday in our NMLRG #3, Athens,&nbsp;we did discussed this&nb=
sp;issue regarding to the potential standardized dataset. And we are planni=
ng to have an open discussion to regarding
 to the requirements and validation qualification of the potential common d=
ataset in the coming NMLRG #4, Berlin, IETF96.</p>
<p>&nbsp;</p>
<p>However, upto now, I have serveral concerns. Firstly, before we could di=
scuss such dataset, we need to narrow our target scenarios to a few very&nb=
sp;specific use cases, giving that there are so many various scenarios in t=
he network area and they are so different
 from each other and require different dataset. Secondly, as we did talk wi=
th operators, the real data is the important properties for them. So, it wo=
uld be almost impossible for them to share them. It leaves two possibilitie=
s on how to get dataset: a) generating
 dataset using simulation; b) try to get some partial dataset from operator=
s that have removed their sensenble data. Both way are problemstic, in the =
risk that a lot of efforts may waste in wrong areas. Thirdly, the availabil=
ity of data&nbsp;may very different in
 different network environments. It depends on the measurement/congnitive f=
unctions that have been deployed or implemented. &nbsp;&nbsp;</p>
<p>&nbsp;</p>
<p>Although having the above concerns, I &nbsp;think we should: a) research=
 on certain dataset. It&nbsp;would be very useful to prove the machine lear=
ning could perform better or provide more flexibility and adaptibility; b) =
continue our discuss on various network use
 cases that could benefit from applying machine learning, in order to serve=
 wider network communities.</p>
<p>&nbsp;</p>
<p>Another important point I would like to discuss is that you mentioned &q=
uot;our goal is to provide accurate, repeatable, and explainable results.&q=
uot; Ideally, we do want this. It is our traditional way for logic-based pr=
ogramming. However, I am not sure that is
 a doable target&nbsp;with machine learning. Many of machine learning algor=
ithms&nbsp;are based on statistics theory, which may not be that much accur=
ate. Some of machine learning algorithms&nbsp;are black box or have very co=
mplexicated internal computing process. So, the
 resulf of these algorithms&nbsp;may be hard to be explained. Also the proc=
ess of machine learning may be hard to be controlled or intervened. Repeata=
ble, yes, we have to have this feature to be widely adopted. However, repea=
table may have to come with some level
 of toleration for the inaccurate.</p>
<p>&nbsp;</p>
<p>Best regards,</p>
<p>&nbsp;</p>
<p>Sheng</p>
<p>&nbsp;</p>
<p></p>
<hr tabindex=3D"-1">
<p></p>
<div style=3D"FONT-FAMILY: Times New Roman; COLOR: #000000; FONT-SIZE: 16px=
">
<div style=3D"DIRECTION: ltr" id=3D"divRpF678056"><font color=3D"#000000" s=
ize=3D"2" face=3D"Tahoma"><b>From:</b> David Meyer [dmm@1-4-5.net]<br>
<b>Sent:</b> 28 June 2016 23:30<br>
<b>To:</b> Sheng Jiang<br>
<b>Cc:</b> nmlrg@irtf.org<br>
<b>Subject:</b> Re: [Nmlrg] links to each slide of presentations//RE: slide=
s of NMLRG #3, June 27th, Athens<br>
</font><br>
</div>
<div></div>
<div>
<div dir=3D"ltr">Sheng,
<div><br>
</div>
<div>Thanks for the pointers. One thing I notice is that we don't have much=
 on what the characteristics of network data might be, and as such, what ki=
nds of existing learning algorithms might be suitable and where additional =
development might be required. For
 example, is the collected data IID? Is it time series? Is the underlying d=
istribution stationary? If not, how does that constrain algorithms we might=
 use or develop? For example, how does a given algorithm deal with concept =
drift/internal covariate shift (in
 the case of DNNs; see e.g., batch normalization [0]). There are many other=
 such questions, such as is the data categorical (e.g., ports, IP addresses=
) or is it&nbsp;continuous/discrete (e.g., counters). And if the data in qu=
estion is categorical, what is the&nbsp;cardinality
 of the categories (this will inform how such data can be encoded); in the =
case of IP addresses we can't really one-hot encode addresses because their=
 cardinality is too large (2^32 or 2^128); this has implications for how we=
 build classifiers (in particular,
 for softmax layers in DNNs of various kinds).</div>
<div><br>
</div>
<div>Related to the above is the question of features? What are good featur=
es for networking? Where do they come from? Are they domain specific? Can w=
e learn features the way a DNNs does in the network space? Can we use autoe=
ncoders to discover such features?
 Or can we use GANs to train DNNs for network classification tasks in an un=
supervised manner? Are there other, non-ad-hoc (well founded) methods we ca=
n use, or is every use case a one-off (one would hope not).</div>
<div><br>
</div>
<div>We can carry the same kinds of analyzes to the algorithms applied. For=
 example, while something k-means is an effective way to get a feeling for =
how&nbsp;continuous/discrete hangs together, if our data is categorical sta=
tistical clustering approaches such as
 LDA might provide a more well-founded approach (of course, as with most Ba=
yesian techniques is the question of approximate inference arises since in =
most interesting cases the integral that we need to solve, namely the margi=
nal probability data isn't tractable
 so we need to resort to MCMC or more likely&nbsp;variational inference). A=
nd what about the use of SGD/batch normalization etc with DNNs, and perhaps=
 more importantly, can we use network data to train DNN policy networks for=
 reinforcement learning like we saw in
 deep Q-learning and AlphaGo?</div>
<div><br>
</div>
<div>These comments are all by way of saying that we don't have a solid the=
oretical understanding (yet) of how techniques that have been so successful=
 in other domains (e.g., DNNs for perceptual tasks) generalize to networkin=
g use cases. We will need this understanding
 if our goal is to provide accurate, repeatable, and explainable results.&n=
bsp;</div>
<div><br>
</div>
<div>In order to accomplish all of this we need, as I have been saying , no=
t only a good understanding of how these algorithms work but also standardi=
zed data sets and associated benchmarks so we can tell if we are making pro=
gress (or even if our techniques
 work). Analogies here include MNIST and ImageNet and their associated benc=
hmarks, among others. As mentioned &nbsp;standardized data sets are key to =
making progress in the ML for networking space (otherwise how do you know y=
our technique works and/or improves on
 another techniques?). One might assume that these data sets would need to =
be labeled (as supervised learning is where most of the progress is being m=
ade these days), but not necessarily; Generative Adversarial Networks (GANs=
) have emerged as a new way to train
 DNNs in an unsupervised manner (this is moving very rapidly; see e.g.,&nbs=
p;<a href=3D"https://openai.com/blog/generative-models/" target=3D"_blank">=
https://openai.com/blog/generative-models/</a>).</div>
<div><br>
</div>
<div>The summary here is that the &quot;distance&quot; between theory and p=
ractice in ML is effectively zero right now due to the incredible rate of p=
rogress in the field; this means we need &nbsp;to understand both sides of =
the theory/practice coin in order to be effective.
 None of the slide decks provide much background on what the proposed algor=
ithms are, how they work, or why they should be expected to work on network=
 data.</div>
<div><br>
</div>
<div>Finally, if you are interested in LDA or other algorithms there are a =
few short explanatory pieces I have written for my team on
<a href=3D"http://www.1-4-5.net/~dmm/ml" target=3D"_blank">http://www.1-4-5=
.net/~dmm/ml</a> (works in progress).</div>
<div><br>
</div>
<div>Thanks,</div>
<div><br>
Dave</div>
<div><br>
</div>
<div>[0]&nbsp;<a href=3D"https://arxiv.org/pdf/1502.03167.pdf" target=3D"_b=
lank">https://arxiv.org/pdf/1502.03167.pdf</a></div>
</div>
<div class=3D"gmail_extra"><br>
<div class=3D"gmail_quote">On Tue, Jun 28, 2016 at 7:03 AM, Sheng Jiang <sp=
an dir=3D"ltr">
&lt;<a href=3D"mailto:jiangsheng@huawei.com" target=3D"_blank">jiangsheng@h=
uawei.com</a>&gt;</span> wrote:<br>
<blockquote style=3D"BORDER-LEFT: #ccc 1px solid; MARGIN: 0px 0px 0px 0.8ex=
; PADDING-LEFT: 1ex" class=3D"gmail_quote">
Oops... The proceeding page for interim meeting seems not be as intelligent=
 as the proceeding pages of IETF meetings. Our proceeding page does not aut=
onomically show the slides. I have sent an email to IETF secretary to ask t=
hem to fix it. Meanwhile, in this
 email, here is the links for each presentations:<br>
<br>
Slide Filename Edit Replace Delete<br>
Chair Slides<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-0.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-0.pdf</a><br>
<br>
Introduction to Network Machine Learning &amp; NMLRG<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-1.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-1.pdf</a><br>
<br>
Data Collection and Analysis At High Security Lab<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-2.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-2.pdf</a><br>
<br>
Use Cases of Applying Machine Learning Mechanism with Network Traffic<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-3.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-3.pdf</a><br>
<br>
Mobile network state characterization and prediction<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-4.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-4.pdf</a><br>
<br>
Learning how to route<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-5.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-5.pdf</a><br>
<br>
Regards,<br>
<br>
Sheng<br>
________________________________________<br>
From: nmlrg [<a href=3D"mailto:nmlrg-bounces@irtf.org" target=3D"_blank">nm=
lrg-bounces@irtf.org</a>] on behalf of Sheng Jiang [<a href=3D"mailto:jiang=
sheng@huawei.com" target=3D"_blank">jiangsheng@huawei.com</a>]<br>
Sent: 28 June 2016 21:03<br>
To: <a href=3D"mailto:nmlrg@irtf.org" target=3D"_blank">nmlrg@irtf.org</a><=
br>
Subject: [Nmlrg] slides of NMLRG #3, June 27th, Athens<br>
<br>
Hi, nmlrg,<br>
<br>
All slides that have been presented in our NMLRG #3 meeting, June 27th, 201=
6, Athens, Greece, with EUCNC2016, have been uploaded. They can be accessed=
 through below link<br>
<br>
<a href=3D"https://www.ietf.org/proceedings/interim/2016/06/27/nmlrg/procee=
dings.html" rel=3D"noreferrer" target=3D"_blank">https://www.ietf.org/proce=
edings/interim/2016/06/27/nmlrg/proceedings.html</a><br>
<br>
Best regards,<br>
<br>
Sheng<br>
_______________________________________________<br>
nmlrg mailing list<br>
<a href=3D"mailto:nmlrg@irtf.org" target=3D"_blank">nmlrg@irtf.org</a><br>
<a href=3D"https://www.irtf.org/mailman/listinfo/nmlrg" rel=3D"noreferrer" =
target=3D"_blank">https://www.irtf.org/mailman/listinfo/nmlrg</a><br>
_______________________________________________<br>
nmlrg mailing list<br>
<a href=3D"mailto:nmlrg@irtf.org" target=3D"_blank">nmlrg@irtf.org</a><br>
<a href=3D"https://www.irtf.org/mailman/listinfo/nmlrg" rel=3D"noreferrer" =
target=3D"_blank">https://www.irtf.org/mailman/listinfo/nmlrg</a><br>
</blockquote>
</div>
<br>
</div>
</div>
</div>
</div>
</body>
</html>

--_000_5D36713D8A4E7348A7E10DF7437A4B927CA897D4NKGEML515MBXchi_--


From nobody Wed Jun 29 08:35:46 2016
Return-Path: <dmm@1-4-5.net>
X-Original-To: nmlrg@ietfa.amsl.com
Delivered-To: nmlrg@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 30A5C12D965 for <nmlrg@ietfa.amsl.com>; Wed, 29 Jun 2016 08:35:45 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.599
X-Spam-Level: 
X-Spam-Status: No, score=-2.599 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, HTML_MESSAGE=0.001, RCVD_IN_DNSWL_LOW=-0.7] autolearn=ham autolearn_force=no
Authentication-Results: ietfa.amsl.com (amavisd-new); dkim=pass (2048-bit key) header.d=1-4-5-net.20150623.gappssmtp.com
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id RAS8i4IvIdYA for <nmlrg@ietfa.amsl.com>; Wed, 29 Jun 2016 08:35:40 -0700 (PDT)
Received: from mail-it0-x22f.google.com (mail-it0-x22f.google.com [IPv6:2607:f8b0:4001:c0b::22f]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by ietfa.amsl.com (Postfix) with ESMTPS id 2552312D0B7 for <nmlrg@irtf.org>; Wed, 29 Jun 2016 08:35:39 -0700 (PDT)
Received: by mail-it0-x22f.google.com with SMTP id h190so45027886ith.1 for <nmlrg@irtf.org>; Wed, 29 Jun 2016 08:35:39 -0700 (PDT)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1-4-5-net.20150623.gappssmtp.com; s=20150623; h=mime-version:in-reply-to:references:from:date:message-id:subject:to :cc; bh=FZ/GkN9RfkArqMmnEDDU6Pb70vTmFijnZX0xBGkIH7M=; b=Mwe8AxiaPY+SFTuZ5ODXE9x88qqWAfT8XqjqQHubzeNImZDRQAQ0VsoQIaEGhwRm5x DM9VdMCoPp7QKNHYqqFb+roXSz61YRB2EhNsMfWp8OuaRqcUNfZZCUiebSnL2YrUoFfL XySnbcjMyHCdccwO3OfXbcAFDQhJPMjFU+e5Y2oYQygrGphdUbF9Nduv0guy1Z5vslVu JwZx95R/LxX39O0f5OGcNDBbLUM+1YiTqJuNJxnWbIscrqXQtGe0/9Q/GhJ9XrGIIs7s gwRjigcY2gqGJnSrgOGQxMr6e4zW3mPmetWd4hLHbdkqbCRnWOi1vx6B3vpJ83ruB8GB BN5w==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:from:date :message-id:subject:to:cc; bh=FZ/GkN9RfkArqMmnEDDU6Pb70vTmFijnZX0xBGkIH7M=; b=daUWHAGrFnIdCUuK4dyvfSJ1rEw8frjEiTIIIZbN7f5PLnqg7KR7ZlX6/SVhksVgKc qzLQOkQD9XbvrdrnEbdxW7gDltymJSsHpHMsPfewpcZlPq7bj5+X/NpctH4VuxNdZG58 F+rKIgB+PxRWpowhYPHB90AnbU45ttLSxa57ttiXHDaevp0ig2z2ntUKbnAFKx1EHLzP 9T0HkNzUMg/nASkQladyiL7YS/nhUIqpsOTID14Mr/EJCmnQCoLK5bEGUnMDUTKyursn /J0hGeg6H12lR5MG8ACzYNSaOBWtS2d8UFKzu0ZzbVToDqaiamprNq3V9W8EoE1RKtuk LFNA==
X-Gm-Message-State: ALyK8tIz46xjtYzdJR9D29I3CA9sYtzAwCNymGDb6EoDNldzOUFJmVwin0Fdh80gGV3HPb2H+Vzc7KqlfFRLMg==
X-Received: by 10.36.61.204 with SMTP id n195mr22079306itn.92.1467214538135; Wed, 29 Jun 2016 08:35:38 -0700 (PDT)
MIME-Version: 1.0
Received: by 10.64.148.133 with HTTP; Wed, 29 Jun 2016 08:35:37 -0700 (PDT)
X-Originating-IP: [128.223.156.253]
In-Reply-To: <5D36713D8A4E7348A7E10DF7437A4B927CA897D4@NKGEML515-MBX.china.huawei.com>
References: <5D36713D8A4E7348A7E10DF7437A4B927CA893E9@NKGEML515-MBX.china.huawei.com> <CAHiKxWhfdukjRVnSKnbLwMwamiqZYoCD350BoNqO4njqkQoU1g@mail.gmail.com> <5D36713D8A4E7348A7E10DF7437A4B927CA897D4@NKGEML515-MBX.china.huawei.com>
From: David Meyer <dmm@1-4-5.net>
Date: Wed, 29 Jun 2016 08:35:37 -0700
Message-ID: <CAHiKxWhMPjKY24zucg-UND0ARWHbjnOyYv=Tia-yk-Vq6Ta_SA@mail.gmail.com>
To: Sheng Jiang <jiangsheng@huawei.com>
Content-Type: multipart/alternative; boundary=001a1144404e5b125405366c811f
Archived-At: <https://mailarchive.ietf.org/arch/msg/nmlrg/JXzVLU6oO4naytteFVUufgw0fg0>
Cc: "nmlrg@irtf.org" <nmlrg@irtf.org>
Subject: Re: [Nmlrg] links to each slide of presentations//RE: slides of NMLRG #3, June 27th, Athens
X-BeenThere: nmlrg@irtf.org
X-Mailman-Version: 2.1.17
Precedence: list
List-Id: Network Machine Learning Research Group <nmlrg.irtf.org>
List-Unsubscribe: <https://www.irtf.org/mailman/options/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=unsubscribe>
List-Archive: <https://mailarchive.ietf.org/arch/browse/nmlrg/>
List-Post: <mailto:nmlrg@irtf.org>
List-Help: <mailto:nmlrg-request@irtf.org?subject=help>
List-Subscribe: <https://www.irtf.org/mailman/listinfo/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=subscribe>
X-List-Received-Date: Wed, 29 Jun 2016 15:35:45 -0000

--001a1144404e5b125405366c811f
Content-Type: text/plain; charset=UTF-8

On Wed, Jun 29, 2016 at 1:10 AM, Sheng Jiang <jiangsheng@huawei.com> wrote:

> Hi, David,
>
>
>
> Thanks so much for your email. I am sharing the same concern you have.
> Actually, this Monday in our NMLRG #3, Athens, we did discussed this issue
> regarding to the potential standardized dataset. And we are planning to
> have an open discussion to regarding to the requirements and validation
> qualification of the potential common dataset in the coming NMLRG #4,
> Berlin, IETF96.
>

Looking forward to it.


>
>
> However, upto now, I have serveral concerns. Firstly, before we could
> discuss such dataset, we need to narrow our target scenarios to a few
> very specific use cases, giving that there are so many various scenarios in
> the network area and they are so different from each other and require
> different dataset.
>

I'm hoping that not every network is not a one off. Rather, what we seek is
to learn what are the fundamental properties of networks that can be the
basis of reasoning about networks do. For example, we should be able to
learn something from work done on transfer learning [0] and representation
learning [1]. In the image recognition space there are even startups that
provide pretrained AlexNet, GooLeNet, etc trained on ImageNet (and others)
that you can further train/customize for your particular application. In
any event, if every network requires something totally different then we
haven't yet understood what is fundamental to the structure of our
networks.

> Secondly, as we did talk with operators, the real data is the important
> properties for them. So, it would be almost impossible for them to share
> them. It leaves two possibilities on how to get dataset: a) generating
> dataset using simulation;
>

This likely won't work. For example, if you do this then try to train a DNN
on that data set the NN will learn how the simulator works. This is
unlikely to be the outcome that we'd like (we already know how the
simulator works).

 b) try to get some partial dataset from operators that have removed their
sensenble data. Both way are problemstic, in the risk that a lot of efforts
may waste in wrong areas.

Agree that this is problematic.

Thirdly, the availability of data may very different in different network
environments. It depends on the measurement/congnitive functions that have
been deployed or implemented.

What is a "cognitive function"?

In any event, you have outlined just a few of the problems with creating
"standardized" data sets for our use. I have quoted standardized because
I'm beginning to think its the wrong word (invokes "standards" e.g., IETF,
<X>..., and that is not what is meant here).  Not sure what the right term
is but  "standardized" seems to be the wrong word here.


>
> Although having the above concerns, I  think we should: a) research on
> certain dataset. It would be very useful to prove the machine learning
> could perform better or provide more flexibility and adaptibility;
>

Not sure what you mean here. By "certain dataset" you mean what?

> b) continue our discuss on various network use cases that could benefit
> from applying machine learning, in order to serve wider network communities.
>

Use cases are good, however machine learning isn't something we can just
apply to a use case because we've defined the use case. There are really at
least three parts of this that need to be considered as part of a solution:
(i). use case (as you mention), (ii). algorithm(s) that are going to
provide the regression/classification/ desired in (i), and (iii). data
sets.

(ii). requires understanding of the algorithms that can be applied to
produce the inferences (regressions, ...)  we are interested in. What
algorithms we can use depend on the data we can get (iii).

That said, one of the things we've learned over the past decade or so is
that learning itself is a very general process. We see this in how powerful
something as simple as SGD/backprop is (see e.g., [2]). We can also use
this to our advantage and re-purpose algorithms originally developed for
other tasks. A simple example of this is LDA (see [3,4]). Here the original
idea was to model topics in a corpus of documents, where the topics
themselves (along with the per-document topic proportions) were the
latent/hidden variables. One way to re-purpose LDA for applications like
ours (essentially classification/anomaly detection tasks) is to view each
to/from pair of IP addresses as a "document" (in the language of LDA) and
then words are constructed from features of the to/from traffic (e.g., to
ports, etc). See [5] for a decent example of how you might think about this
kind of "statistical clustering" using the LDA Probabilistic Graphical
Model approach. So there is hope, but there is a lot of work left to be
done on both the theory and practice sides of this (which as I mentioned
earlier are, at least currently, essentially the same thing).

>
>
> Another important point I would like to discuss is that you mentioned "our
> goal is to provide accurate, repeatable, and explainable results." Ideally,
> we do want this. It is our traditional way for logic-based programming.
>
What is "logic-based programming"?  Do you mean logic programming, e.g.,
Prolog? If so I understand logic programming (I did a lot of work on logic
programming back in the late 80s/early 90s, see
http://dl.acm.org/citation.cfm?id=87991). But even so, I don't understand
the connection, so can you be more explicit?


> However, I am not sure that is a doable target with machine learning. Many
> of machine learning algorithms are based on statistics theory, which may
> not be that much accurate.
>
Sorry, I don't understand this. Things like Bayesian inference (e.g., LDA)
are inherently about quantifying uncertainty. In any event can you be more
specific? A learning algorithm can "accurately" approximate something like
a posterior distribution, its just that the output is just that, a
posterior distribution. So there is a difference between "accuracy" (which
isn't really defined here) and the stochastic nature of the results we get
from ML.

OTOH, if you mean accuracy of our estimated distributions then there are
obviously ways this can go wrong. For example, even an algorithm as simple
as k-means can go wrong since the optimization objective (distortion
function) is non-convex, so it is sensitive to initial conditions (in
particular, where you start; if you are unlucky you can get stuck in a
local minimum  due to the non-convex nature of the objective). See e.g.,
[8]. BTW,  k-means was called out in one of the decks below but as I
mentioned, not much discussion of the important topics we are discussing
here there. There is also approximate inference (for example, MCMC or
variational approaches), but these approaches also have
provably quantitative error bounds (http://www.1-4-5.net/~dmm/ml/ps.pdf
talks a bit about how variational approaches, essential trading inference
for optimization where the inference (integral) is intractable, find a
tight lower-bound using Jensen's inequality.

> Some of machine learning algorithms are black box or have very
> complexicated internal computing process. So, the resulf of these
> algorithms may be hard to be explained.
>
This is definitely true to some extent today with DNNs, but that is
changing rapidly. Results can be repeatable but explaining exactly what
units (artificial neurons) far from the input are doing can be difficult.
But this too is an area of active research and development (see [6] from
last year's ICML, or the work I pointed to on GANs in my last note as
progress in this area). Other approaches such probabilistic graphical
models or signal processing approaches are perhaps more easily explained,
but I don't think this is the level of explanation that is going to be
useful to people deploying and operating networks, so we'll have to find
different ways of explaining what things like DNNs to in a way that is
consumable by operators, etc.

All this by way of saying that we are gaining better understanding how DNNs
work every day (literally) and as a result  I'm optimistic that we can
build technology that will be accurate (again, stochastic results),
repeatable and explainable.

I'll just note here that if you have a model with literally billions of
parameters being trained by SGD its not that surprising that humans have a
hard time understanding what the computation is. As a concrete example,
DNNs can solve the so-called "selectivity-invariance" problem [7], but we
humans don't really  know how to do, so this is another example.


> Also the process of machine learning may be hard to be controlled or
> intervened.
>

Again, not sure what you mean here. Please give a bit more detail.


> Repeatable, yes, we have to have this feature to be widely adopted.
> However, repeatable may have to come with some level of toleration for the
> inaccurate.
>

All of this is inherently probabilistic. A DNN, for example, can output the
posterior distribution over some hidden variables of interest accurately.
LDA finds the latent "topics" and word distributions, both of which are
multinomals so again we're essentially getting a posterior over the hidden
variables.  Even k-means is  just the EM (Expectation Maximization)
algorithm applied to a particular Naive Bayes model. The same can be said
of almost every learning algorithm you might mention. So "inaccurate" isn't
the correct word here (unless you mean whatever you trained didn't predict
whatever you wanted to predict, in which case something else is wrong). So
the DNN gives you as stochastic result, which may be different than what
many network people are used to. But its not inaccurate, its just
stochastic.

Thx,

--dmm



[0] http://people.csail.mit.edu/mcollins/papers/CVPR_2008.pdf
[1]  https://arxiv.org/abs/1206.5538
[2] http://deliprao.com/archives/153
[3] http://www.1-4-5.net/~dmm/ml/lda_intro.pdf
[4] http://www.jmlr.org/papers/volume3/blei03a/blei03a.pdf
[5] http://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=5949386
[6]
http://yosinski.com/media/papers/Yosinski__2015__ICML_DL__Understanding_Neural_Networks_Through_Deep_Visualization__.pdf
[7]
https://cbmm.mit.edu/sites/default/files/publications/Anselmi_Invariance_CBMM_memo_29.pdf
[8] http://cs229.stanford.edu/notes/cs229-notes7a.pdf

> Best regards,
>
>
>
> Sheng
>
>
>
> ------------------------------
>
> *From:* David Meyer [dmm@1-4-5.net]
> *Sent:* 28 June 2016 23:30http://deliprao.com/archives/153
> *To:* Sheng Jiang
> *Cc:* nmlrg@irtf.org
> *Subject:* Re: [Nmlrg] links to each slide of presentations//RE: slides
> of NMLRG #3, June 27th, Athens
>
> Sheng,
>
> Thanks for the pointers. One thing I notice is that we don't have much on
> what the characteristics of network data might be, and as such, what kinds
> of existing learning algorithms might be suitable and where additional
> development might be required. For example, is the collected data IID? Is
> it time series? Is the underlying distribution stationary? If not, how does
> that constrain algorithms we might use or develop? For example, how does a
> given algorithm deal with concept drift/internal covariate shift (in the
> case of DNNs; see e.g., batch normalization [0]). There are many other such
> questions, such as is the data categorical (e.g., ports, IP addresses) or
> is it continuous/discrete (e.g., counters). And if the data in question is
> categorical, what is the cardinality of the categories (this will inform
> how such data can be encoded); in the case of IP addresses we can't really
> one-hot encode addresses because their cardinality is too large (2^32 or
> 2^128); this has implications for how we build classifiers (in particular,
> for softmax layers in DNNs of various kinds).
>
> Related to the above is the question of features? What are good features
> for networking? Where do they come from? Are they domain specific? Can we
> learn features the way a DNNs does in the network space? Can we use
> autoencoders to discover such features? Or can we use GANs to train DNNs
> for network classification tasks in an unsupervised manner? Are there
> other, non-ad-hoc (well founded) methods we can use, or is every use case a
> one-off (one would hope not).
>
> We can carry the same kinds of analyzes to the algorithms applied. For
> example, while something k-means is an effective way to get a feeling for
> how continuous/discrete hangs together, if our data is categorical
> statistical clustering approaches such as LDA might provide a more
> well-founded approach (of course, as with most Bayesian techniques is the
> question of approximate inference arises since in most interesting cases
> the integral that we need to solve, namely the marginal probability data
> isn't tractable so we need to resort to MCMC or more likely variational
> inference). And what about the use of SGD/batch normalization etc with
> DNNs, and perhaps more importantly, can we use network data to train DNN
> policy networks for reinforcement learning like we saw in deep Q-learning
> and AlphaGo?
>
> These comments are all by way of saying that we don't have a solid
> theoretical understanding (yet) of how techniques that have been so
> successful in other domains (e.g., DNNs for perceptual tasks) generalize to
> networking use cases. We will need this understanding if our goal is to
> provide accurate, repeatable, and explainable results.
>
> In order to accomplish all of this we need, as I have been saying , not
> only a good understanding of how these algorithms work but also
> standardized data sets and associated benchmarks so we can tell if we are
> making progress (or even if our techniques work). Analogies here include
> MNIST and ImageNet and their associated benchmarks, among others. As
> mentioned  standardized data sets are key to making progress in the ML for
> networking space (otherwise how do you know your technique works and/or
> improves on another techniques?). One might assume that these data sets
> would need to be labeled (as supervised learning is where most of the
> progress is being made these days), but not necessarily; Generative
> Adversarial Networks (GANs) have emerged as a new way to train DNNs in an
> unsupervised manner (this is moving very rapidly; see e.g.,
> https://openai.com/blog/generative-models/).
>
> The summary here is that the "distance" between theory and practice in ML
> is effectively zero right now due to the incredible rate of progress in the
> field; this means we need  to understand both sides of the theory/practice
> coin in order to be effective. None of the slide decks provide much
> background on what the proposed algorithms are, how they work, or why they
> should be expected to work on network data.
>
> Finally, if you are interested in LDA or other algorithms there are a few
> short explanatory pieces I have written for my team on
> http://www.1-4-5.net/~dmm/ml (works in progress).
>
> Thanks,
>
> Dave
>
> [0] https://arxiv.org/pdf/1502.03167.pdf
>
> On Tue, Jun 28, 2016 at 7:03 AM, Sheng Jiang <jiangsheng@huawei.com>
> wrote:
>
>> Oops... The proceeding page for interim meeting seems not be as
>> intelligent as the proceeding pages of IETF meetings. Our proceeding page
>> does not autonomically show the slides. I have sent an email to IETF
>> secretary to ask them to fix it. Meanwhile, in this email, here is the
>> links for each presentations:
>>
>> Slide Filename Edit Replace Delete
>> Chair Slides
>>
>> https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-0.pdf
>>
>> Introduction to Network Machine Learning & NMLRG
>>
>> https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-1.pdf
>>
>> Data Collection and Analysis At High Security Lab
>>
>> https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-2.pdf
>>
>> Use Cases of Applying Machine Learning Mechanism with Network Traffic
>>
>> https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-3.pdf
>>
>> Mobile network state characterization and prediction
>>
>> https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-4.pdf
>>
>> Learning how to route
>>
>> https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-5.pdf
>>
>> Regards,
>>
>> Sheng
>> ________________________________________
>> From: nmlrg [nmlrg-bounces@irtf.org] on behalf of Sheng Jiang [
>> jiangsheng@huawei.com]
>> Sent: 28 June 2016 21:03
>> To: nmlrg@irtf.org
>> Subject: [Nmlrg] slides of NMLRG #3, June 27th, Athens
>>
>> Hi, nmlrg,
>>
>> All slides that have been presented in our NMLRG #3 meeting, June 27th,
>> 2016, Athens, Greece, with EUCNC2016, have been uploaded. They can be
>> accessed through below link
>>
>> https://www.ietf.org/proceedings/interim/2016/06/27/nmlrg/proceedings.html
>>
>> Best regards,
>>
>> Sheng
>> _______________________________________________
>> nmlrg mailing list
>> nmlrg@irtf.org
>> https://www.irtf.org/mailman/listinfo/nmlrg
>> _______________________________________________
>> nmlrg mailing list
>> nmlrg@irtf.org
>> https://www.irtf.org/mailman/listinfo/nmlrg
>>
>
>

--001a1144404e5b125405366c811f
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><br><div class=3D"gmail_extra"><br><div class=3D"gmail_quo=
te">On Wed, Jun 29, 2016 at 1:10 AM, Sheng Jiang <span dir=3D"ltr">&lt;<a h=
ref=3D"mailto:jiangsheng@huawei.com" target=3D"_blank">jiangsheng@huawei.co=
m</a>&gt;</span> wrote:<br><blockquote class=3D"gmail_quote" style=3D"margi=
n:0px 0px 0px 0.8ex;border-left-width:1px;border-left-style:solid;border-le=
ft-color:rgb(204,204,204);padding-left:1ex">




<div>
<div style=3D"direction:ltr;font-family:Tahoma;color:rgb(0,0,0);font-size:1=
0pt">
<p>Hi, David,</p>
<p>=C2=A0</p>
<p>Thanks so much for your email. I am sharing the same concern you have. A=
ctually, this Monday in our NMLRG #3, Athens,=C2=A0we did discussed this=C2=
=A0issue regarding to the potential standardized dataset. And we are planni=
ng to have an open discussion to regarding
 to the requirements and validation qualification of the potential common d=
ataset in the coming NMLRG #4, Berlin, IETF96.</p></div></div></blockquote>=
<div><br></div><div>Looking forward to it.</div><div>=C2=A0</div><blockquot=
e class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-left-width=
:1px;border-left-style:solid;border-left-color:rgb(204,204,204);padding-lef=
t:1ex"><div><div style=3D"direction:ltr;font-family:Tahoma;color:rgb(0,0,0)=
;font-size:10pt">
<p>=C2=A0</p>
<p>However, upto now, I have serveral concerns. Firstly, before we could di=
scuss such dataset, we need to narrow our target scenarios to a few very=C2=
=A0specific use cases, giving that there are so many various scenarios in t=
he network area and they are so different
 from each other and require different dataset. </p></div></div></blockquot=
e><div><br></div><div>I&#39;m hoping that not every network is not a one of=
f. Rather, what we seek is to learn what are the fundamental properties of =
networks that can be the basis of reasoning about networks do. For example,=
 we should be able to learn something from work done on transfer learning [=
0] and representation learning [1]. In the image recognition space there ar=
e even startups that provide pretrained AlexNet, GooLeNet, etc trained on I=
mageNet (and others) that you can further train/customize for your particul=
ar application. In any event, if every network requires something totally d=
ifferent then we haven&#39;t yet understood what is fundamental to the stru=
cture of our networks.=C2=A0</div><blockquote class=3D"gmail_quote" style=
=3D"margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-style:solid;=
border-left-color:rgb(204,204,204);padding-left:1ex"><div><div style=3D"dir=
ection:ltr;font-family:Tahoma;color:rgb(0,0,0);font-size:10pt"><p>Secondly,=
 as we did talk with operators, the real data is the important properties f=
or them. So, it would be almost impossible for them to share them. It leave=
s two possibilities on how to get dataset: a) generating
 dataset using simulation;</p></div></div></blockquote><div><br></div><div>=
This likely won&#39;t work. For example, if you do this then try to train a=
 DNN on that data set the NN will learn how the simulator works. This is un=
likely to be the outcome that we&#39;d like (we already know how the simula=
tor works).</div><div><br></div><div>=C2=A0<span style=3D"color:rgb(0,0,0);=
font-family:Tahoma;font-size:10pt">b) try to get some partial dataset from =
operators that have removed their sensenble data. Both way are problemstic,=
 in the risk that a lot of efforts may waste in wrong areas.=C2=A0</span></=
div><div><span style=3D"color:rgb(0,0,0);font-family:Tahoma;font-size:10pt"=
><br></span></div><div><span style=3D"color:rgb(0,0,0);font-family:Tahoma;f=
ont-size:10pt">Agree that this is problematic.=C2=A0</span></div><div><span=
 style=3D"color:rgb(0,0,0);font-family:Tahoma;font-size:10pt"><br></span></=
div><div><span style=3D"color:rgb(0,0,0);font-family:Tahoma;font-size:10pt"=
>Thirdly, the availability of data=C2=A0may very different in
 different network environments. It depends on the measurement/congnitive f=
unctions that have been deployed or implemented. =C2=A0=C2=A0</span></div><=
div><span style=3D"color:rgb(0,0,0);font-family:Tahoma;font-size:10pt"><br>=
</span></div><div><span style=3D"color:rgb(0,0,0);font-family:Tahoma;font-s=
ize:10pt">What is a &quot;cognitive function&quot;?</span></div><div><span =
style=3D"color:rgb(0,0,0);font-family:Tahoma;font-size:10pt"><br></span></d=
iv><div><span style=3D"color:rgb(0,0,0);font-family:Tahoma;font-size:10pt">=
In any event, you have outlined just a few of the problems with creating &q=
uot;standardized&quot; data sets for our use. I have quoted standardized be=
cause I&#39;m beginning to think its the wrong word (invokes &quot;standard=
s&quot; e.g., IETF, &lt;X&gt;..., and that is not what is meant here).=C2=
=A0 Not sure what the right term is but =C2=A0&quot;standardized&quot; seem=
s to be the wrong word here.</span></div><div><span style=3D"color:rgb(0,0,=
0);font-family:Tahoma;font-size:10pt"><br></span></div><blockquote class=3D=
"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-left-width:1px;borde=
r-left-style:solid;border-left-color:rgb(204,204,204);padding-left:1ex"><di=
v style=3D"direction:ltr;font-family:Tahoma;color:rgb(0,0,0);font-size:10pt=
">
<p>=C2=A0</p>
<p>Although having the above concerns, I =C2=A0think we should: a) research=
 on certain dataset. It=C2=A0would be very useful to prove the machine lear=
ning could perform better or provide more flexibility and adaptibility; </p=
></div></blockquote><div><br></div><div>Not sure what you mean here. By &qu=
ot;certain dataset&quot; you mean what?=C2=A0</div><blockquote class=3D"gma=
il_quote" style=3D"margin:0px 0px 0px 0.8ex;border-left-width:1px;border-le=
ft-style:solid;border-left-color:rgb(204,204,204);padding-left:1ex"><div st=
yle=3D"direction:ltr;font-family:Tahoma;color:rgb(0,0,0);font-size:10pt"><p=
>b) continue our discuss on various network use
 cases that could benefit from applying machine learning, in order to serve=
 wider network communities.</p></div></blockquote><div><br></div><div>Use c=
ases are good, however machine learning isn&#39;t something we can just app=
ly to a use case because we&#39;ve defined the use case. There are really a=
t least three parts of this that need to be considered as part of a solutio=
n: (i). use case (as you mention), (ii). algorithm(s) that are going to pro=
vide the regression/classification/ desired in (i), and (iii). data sets.=
=C2=A0</div><div><br></div><div>(ii). requires understanding of the algorit=
hms that can be applied to produce the inferences (regressions, ...) =C2=A0=
we are interested in. What algorithms we can use depend on the data we can =
get (iii). =C2=A0</div><div><br></div><div>That said, one of the things we&=
#39;ve learned over the past decade or so is that learning itself is a very=
 general process. We see this in how powerful something as simple as SGD/ba=
ckprop is (see e.g., [2]). We can also use this to our advantage and re-pur=
pose algorithms originally developed for other tasks. A simple example of t=
his is LDA (see [3,4]). Here the original idea was to model topics in a cor=
pus of documents, where the topics themselves (along with the per-document =
topic proportions) were the latent/hidden variables. One way to re-purpose =
LDA for applications like ours (essentially classification/anomaly detectio=
n tasks) is to view each to/from pair of IP addresses as a &quot;document&q=
uot; (in the language of LDA) and then words are constructed from features =
of the to/from traffic (e.g., to ports, etc). See [5] for a decent example =
of how you might think about this kind of &quot;statistical clustering&quot=
; using the LDA Probabilistic Graphical Model approach. So there is hope, b=
ut there is a lot of work left to be done on both the theory and practice s=
ides of this (which as I mentioned earlier are, at least currently, essenti=
ally the same thing).</div><blockquote class=3D"gmail_quote" style=3D"margi=
n:0px 0px 0px 0.8ex;border-left-width:1px;border-left-style:solid;border-le=
ft-color:rgb(204,204,204);padding-left:1ex"><div style=3D"direction:ltr;fon=
t-family:Tahoma;color:rgb(0,0,0);font-size:10pt">
<p>=C2=A0</p>
<p>Another important point I would like to discuss is that you mentioned &q=
uot;our goal is to provide accurate, repeatable, and explainable results.&q=
uot; Ideally, we do want this. It is our traditional way for logic-based pr=
ogramming.</p></div></blockquote><div>What is &quot;logic-based programming=
&quot;?=C2=A0 Do you mean logic programming, e.g., Prolog? If so I understa=
nd logic programming (I did a lot of work on logic programming back in the =
late 80s/early 90s, see=C2=A0<a href=3D"http://dl.acm.org/citation.cfm?id=
=3D87991">http://dl.acm.org/citation.cfm?id=3D87991</a>). But even so, I do=
n&#39;t understand the connection, so can you be more explicit?</div><div>=
=C2=A0</div><blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0=
.8ex;border-left-width:1px;border-left-style:solid;border-left-color:rgb(20=
4,204,204);padding-left:1ex"><div style=3D"direction:ltr;font-family:Tahoma=
;color:rgb(0,0,0);font-size:10pt"><p>However, I am not sure that is
 a doable target=C2=A0with machine learning. Many of machine learning algor=
ithms=C2=A0are based on statistics theory, which may not be that much accur=
ate. </p></div></blockquote><div>Sorry, I don&#39;t understand this. Things=
 like Bayesian inference (e.g., LDA) are inherently about quantifying uncer=
tainty. In any event can you be more specific? A learning algorithm can &qu=
ot;accurately&quot; approximate something like a posterior distribution, it=
s just that the output is just that, a posterior distribution. So there is =
a difference between &quot;accuracy&quot; (which isn&#39;t really defined h=
ere) and the stochastic nature of the results we get from ML.</div><div><br=
></div><div>OTOH, if you mean accuracy of our estimated distributions then =
there are obviously ways this can go wrong. For example, even an algorithm =
as simple as k-means can go wrong since the optimization objective (distort=
ion function) is non-convex, so it is sensitive to initial conditions (in p=
articular, where you start; if you are unlucky you can get stuck in a local=
 minimum =C2=A0due to the non-convex nature of the objective). See e.g., [8=
]. BTW, =C2=A0k-means was called out in one of the decks below but as I men=
tioned, not much discussion of the important topics we are discussing here =
there. There is also approximate inference (for example, MCMC or variationa=
l approaches), but these approaches also have provably=C2=A0quantitative er=
ror bounds (<a href=3D"http://www.1-4-5.net/~dmm/ml/ps.pdf">http://www.1-4-=
5.net/~dmm/ml/ps.pdf</a> talks a bit about how variational approaches, esse=
ntial trading inference for optimization where the inference (integral) is =
intractable, find a tight lower-bound using Jensen&#39;s inequality.=C2=A0<=
/div><blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;bo=
rder-left-width:1px;border-left-style:solid;border-left-color:rgb(204,204,2=
04);padding-left:1ex"><div style=3D"direction:ltr;font-family:Tahoma;color:=
rgb(0,0,0);font-size:10pt"><p>Some of machine learning algorithms=C2=A0are =
black box or have very complexicated internal computing process. So, the
 resulf of these algorithms=C2=A0may be hard to be explained. </p></div></b=
lockquote><div>This is definitely true to some extent today with DNNs, but =
that is changing rapidly. Results can be repeatable but explaining exactly =
what units (artificial neurons) far from the input are doing can be difficu=
lt. But this too is an area of active research and development (see [6] fro=
m last year&#39;s ICML, or the work I pointed to on GANs in my last note as=
 progress in this area). Other approaches such probabilistic graphical mode=
ls or signal processing approaches are perhaps more easily explained, but I=
 don&#39;t think this is the level of explanation that is going to be usefu=
l to people deploying and operating networks, so we&#39;ll have to find dif=
ferent ways of explaining what things like DNNs to in a way that is consuma=
ble by operators, etc. =C2=A0</div><div><br></div><div>All this by way of s=
aying that we are gaining better understanding how DNNs work every day (lit=
erally) and as a result =C2=A0I&#39;m optimistic that we can build technolo=
gy that will be accurate (again, stochastic results), repeatable and explai=
nable.=C2=A0</div><div><br></div><div>I&#39;ll just note here that if you h=
ave a model with literally billions of parameters being trained by SGD its =
not that surprising that humans have a hard time understanding what the com=
putation is. As a concrete example, DNNs can solve the so-called &quot;sele=
ctivity-invariance&quot; problem [7], but we humans don&#39;t really =C2=A0=
know how to do, so this is another example.=C2=A0</div><div>=C2=A0</div><bl=
ockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-lef=
t-width:1px;border-left-style:solid;border-left-color:rgb(204,204,204);padd=
ing-left:1ex"><div style=3D"direction:ltr;font-family:Tahoma;color:rgb(0,0,=
0);font-size:10pt"><p>Also the process of machine learning may be hard to b=
e controlled or intervened. </p></div></blockquote><div><br></div><div>Agai=
n, not sure what you mean here. Please give a bit more detail.</div><div>=
=C2=A0</div><blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0=
.8ex;border-left-width:1px;border-left-style:solid;border-left-color:rgb(20=
4,204,204);padding-left:1ex"><div style=3D"direction:ltr;font-family:Tahoma=
;color:rgb(0,0,0);font-size:10pt"><p>Repeatable, yes, we have to have this =
feature to be widely adopted. However, repeatable may have to come with som=
e level
 of toleration for the inaccurate.</p></div></blockquote><div><br></div><di=
v>All of this is inherently probabilistic. A DNN, for example, can output t=
he posterior distribution over some hidden variables of interest accurately=
. LDA finds the latent &quot;topics&quot; and word distributions, both of w=
hich are multinomals so again we&#39;re essentially getting a posterior ove=
r the hidden variables.=C2=A0 Even k-means is =C2=A0just the EM (Expectatio=
n Maximization) algorithm applied to a particular Naive Bayes model. The sa=
me can be said of almost every learning algorithm you might mention. So &qu=
ot;inaccurate&quot; isn&#39;t the correct word here (unless you mean whatev=
er you trained didn&#39;t predict whatever you wanted to predict, in which =
case something else is wrong). So the DNN gives you as stochastic result, w=
hich may be different than what many network people are used to. But its no=
t inaccurate, its just stochastic.</div><div><br></div><div>Thx,</div><div>=
<br></div><div>--dmm</div><div><br></div><div><br></div><div><br></div><div=
>[0]=C2=A0<a href=3D"http://people.csail.mit.edu/mcollins/papers/CVPR_2008.=
pdf">http://people.csail.mit.edu/mcollins/papers/CVPR_2008.pdf</a></div><di=
v>[1] =C2=A0<span style=3D"color:rgb(0,0,0);font-family:Tahoma;font-size:10=
pt"><a href=3D"https://arxiv.org/abs/1206.5538">https://arxiv.org/abs/1206.=
5538</a></span></div><div><span style=3D"color:rgb(0,0,0);font-family:Tahom=
a;font-size:10pt">[2]=C2=A0</span><font color=3D"#000000" face=3D"Tahoma"><=
span style=3D"font-size:13.3333px"><a href=3D"http://deliprao.com/archives/=
153">http://deliprao.com/archives/153</a></span></font></div><div><font col=
or=3D"#000000" face=3D"Tahoma"><span style=3D"font-size:13.3333px">[3]=C2=
=A0<a href=3D"http://www.1-4-5.net/~dmm/ml/lda_intro.pdf">http://www.1-4-5.=
net/~dmm/ml/lda_intro.pdf</a></span></font></div><div><font color=3D"#00000=
0" face=3D"Tahoma"><span style=3D"font-size:13.3333px">[4]=C2=A0<a href=3D"=
http://www.jmlr.org/papers/volume3/blei03a/blei03a.pdf">http://www.jmlr.org=
/papers/volume3/blei03a/blei03a.pdf</a></span></font></div><div><font color=
=3D"#000000" face=3D"Tahoma"><span style=3D"font-size:13.3333px">[5]=C2=A0<=
a href=3D"http://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=3D5949386">ht=
tp://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=3D5949386</a></span></fon=
t></div><div><font color=3D"#000000" face=3D"Tahoma"><span style=3D"font-si=
ze:13.3333px">[6]=C2=A0<a href=3D"http://yosinski.com/media/papers/Yosinski=
__2015__ICML_DL__Understanding_Neural_Networks_Through_Deep_Visualization__=
.pdf">http://yosinski.com/media/papers/Yosinski__2015__ICML_DL__Understandi=
ng_Neural_Networks_Through_Deep_Visualization__.pdf</a></span></font></div>=
<div><font color=3D"#000000" face=3D"Tahoma"><span style=3D"font-size:13.33=
33px">[7]=C2=A0<a href=3D"https://cbmm.mit.edu/sites/default/files/publicat=
ions/Anselmi_Invariance_CBMM_memo_29.pdf">https://cbmm.mit.edu/sites/defaul=
t/files/publications/Anselmi_Invariance_CBMM_memo_29.pdf</a></span></font><=
/div><div><font color=3D"#000000" face=3D"Tahoma"><span style=3D"font-size:=
13.3333px">[8]=C2=A0<a href=3D"http://cs229.stanford.edu/notes/cs229-notes7=
a.pdf">http://cs229.stanford.edu/notes/cs229-notes7a.pdf</a></span></font><=
/div><blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;bo=
rder-left-width:1px;border-left-style:solid;border-left-color:rgb(204,204,2=
04);padding-left:1ex"><div><div style=3D"direction:ltr">
<p style=3D"color:rgb(0,0,0);font-family:Tahoma;font-size:10pt">Best regard=
s,</p>
<p style=3D"color:rgb(0,0,0);font-family:Tahoma;font-size:10pt">=C2=A0</p>
<p style=3D"color:rgb(0,0,0);font-family:Tahoma;font-size:10pt">Sheng</p>
<p style=3D"color:rgb(0,0,0);font-family:Tahoma;font-size:10pt">=C2=A0</p>
<p style=3D"color:rgb(0,0,0);font-family:Tahoma;font-size:10pt"></p>
<hr style=3D"color:rgb(0,0,0);font-family:Tahoma;font-size:10pt">
<p style=3D"color:rgb(0,0,0);font-family:Tahoma;font-size:10pt"></p>
<div>
<div style=3D"direction:ltr"><font color=3D"#000000" size=3D"2" face=3D"Tah=
oma" style=3D""><b style=3D"color:rgb(0,0,0);font-family:&quot;Times New Ro=
man&quot;;font-size:16px">From:</b><font face=3D"Times New Roman"><span sty=
le=3D"font-size:16px"> David Meyer [</span></font><a href=3D"mailto:dmm@1-4=
-5.net" target=3D"_blank" style=3D"color:rgb(0,0,0);font-family:&quot;Times=
 New Roman&quot;;font-size:16px">dmm@1-4-5.net</a><font face=3D"Times New R=
oman"><span style=3D"font-size:16px">]</span></font><br>
<b style=3D"color:rgb(0,0,0);font-family:&quot;Times New Roman&quot;;font-s=
ize:16px">Sent:</b><font face=3D"Times New Roman"><span style=3D"font-size:=
16px"> 28 June 2016 23:30</span></font><a href=3D"http://deliprao.com/archi=
ves/153">http://deliprao.com/archives/153</a><br>
<b style=3D"color:rgb(0,0,0);font-family:&quot;Times New Roman&quot;;font-s=
ize:16px">To:</b><font face=3D"Times New Roman"><span style=3D"font-size:16=
px"> Sheng Jiang</span></font><br>
<b style=3D"color:rgb(0,0,0);font-family:&quot;Times New Roman&quot;;font-s=
ize:16px">Cc:</b> <a href=3D"mailto:nmlrg@irtf.org" target=3D"_blank" style=
=3D"color:rgb(0,0,0);font-family:&quot;Times New Roman&quot;;font-size:16px=
">nmlrg@irtf.org</a><br>
<b style=3D"color:rgb(0,0,0);font-family:&quot;Times New Roman&quot;;font-s=
ize:16px">Subject:</b><font face=3D"Times New Roman"><span style=3D"font-si=
ze:16px"> Re: [Nmlrg] links to each slide of presentations//RE: slides of N=
MLRG #3, June 27th, Athens</span></font><br>
</font><br>
</div><div style=3D"color:rgb(0,0,0);font-family:&quot;Times New Roman&quot=
;;font-size:16px"><div class=3D"h5">
<div></div>
<div>
<div dir=3D"ltr">Sheng,
<div><br>
</div>
<div>Thanks for the pointers. One thing I notice is that we don&#39;t have =
much on what the characteristics of network data might be, and as such, wha=
t kinds of existing learning algorithms might be suitable and where additio=
nal development might be required. For
 example, is the collected data IID? Is it time series? Is the underlying d=
istribution stationary? If not, how does that constrain algorithms we might=
 use or develop? For example, how does a given algorithm deal with concept =
drift/internal covariate shift (in
 the case of DNNs; see e.g., batch normalization [0]). There are many other=
 such questions, such as is the data categorical (e.g., ports, IP addresses=
) or is it=C2=A0continuous/discrete (e.g., counters). And if the data in qu=
estion is categorical, what is the=C2=A0cardinality
 of the categories (this will inform how such data can be encoded); in the =
case of IP addresses we can&#39;t really one-hot encode addresses because t=
heir cardinality is too large (2^32 or 2^128); this has implications for ho=
w we build classifiers (in particular,
 for softmax layers in DNNs of various kinds).</div>
<div><br>
</div>
<div>Related to the above is the question of features? What are good featur=
es for networking? Where do they come from? Are they domain specific? Can w=
e learn features the way a DNNs does in the network space? Can we use autoe=
ncoders to discover such features?
 Or can we use GANs to train DNNs for network classification tasks in an un=
supervised manner? Are there other, non-ad-hoc (well founded) methods we ca=
n use, or is every use case a one-off (one would hope not).</div>
<div><br>
</div>
<div>We can carry the same kinds of analyzes to the algorithms applied. For=
 example, while something k-means is an effective way to get a feeling for =
how=C2=A0continuous/discrete hangs together, if our data is categorical sta=
tistical clustering approaches such as
 LDA might provide a more well-founded approach (of course, as with most Ba=
yesian techniques is the question of approximate inference arises since in =
most interesting cases the integral that we need to solve, namely the margi=
nal probability data isn&#39;t tractable
 so we need to resort to MCMC or more likely=C2=A0variational inference). A=
nd what about the use of SGD/batch normalization etc with DNNs, and perhaps=
 more importantly, can we use network data to train DNN policy networks for=
 reinforcement learning like we saw in
 deep Q-learning and AlphaGo?</div>
<div><br>
</div>
<div>These comments are all by way of saying that we don&#39;t have a solid=
 theoretical understanding (yet) of how techniques that have been so succes=
sful in other domains (e.g., DNNs for perceptual tasks) generalize to netwo=
rking use cases. We will need this understanding
 if our goal is to provide accurate, repeatable, and explainable results.=
=C2=A0</div>
<div><br>
</div>
<div>In order to accomplish all of this we need, as I have been saying , no=
t only a good understanding of how these algorithms work but also standardi=
zed data sets and associated benchmarks so we can tell if we are making pro=
gress (or even if our techniques
 work). Analogies here include MNIST and ImageNet and their associated benc=
hmarks, among others. As mentioned =C2=A0standardized data sets are key to =
making progress in the ML for networking space (otherwise how do you know y=
our technique works and/or improves on
 another techniques?). One might assume that these data sets would need to =
be labeled (as supervised learning is where most of the progress is being m=
ade these days), but not necessarily; Generative Adversarial Networks (GANs=
) have emerged as a new way to train
 DNNs in an unsupervised manner (this is moving very rapidly; see e.g.,=C2=
=A0<a href=3D"https://openai.com/blog/generative-models/" target=3D"_blank"=
>https://openai.com/blog/generative-models/</a>).</div>
<div><br>
</div>
<div>The summary here is that the &quot;distance&quot; between theory and p=
ractice in ML is effectively zero right now due to the incredible rate of p=
rogress in the field; this means we need =C2=A0to understand both sides of =
the theory/practice coin in order to be effective.
 None of the slide decks provide much background on what the proposed algor=
ithms are, how they work, or why they should be expected to work on network=
 data.</div>
<div><br>
</div>
<div>Finally, if you are interested in LDA or other algorithms there are a =
few short explanatory pieces I have written for my team on
<a href=3D"http://www.1-4-5.net/~dmm/ml" target=3D"_blank">http://www.1-4-5=
.net/~dmm/ml</a> (works in progress).</div>
<div><br>
</div>
<div>Thanks,</div>
<div><br>
Dave</div>
<div><br>
</div>
<div>[0]=C2=A0<a href=3D"https://arxiv.org/pdf/1502.03167.pdf" target=3D"_b=
lank">https://arxiv.org/pdf/1502.03167.pdf</a></div>
</div>
<div class=3D"gmail_extra"><br>
<div class=3D"gmail_quote">On Tue, Jun 28, 2016 at 7:03 AM, Sheng Jiang <sp=
an dir=3D"ltr">
&lt;<a href=3D"mailto:jiangsheng@huawei.com" target=3D"_blank">jiangsheng@h=
uawei.com</a>&gt;</span> wrote:<br>
<blockquote style=3D"border-left-width:1px;border-left-style:solid;border-l=
eft-color:rgb(204,204,204);margin:0px 0px 0px 0.8ex;padding-left:1ex" class=
=3D"gmail_quote">
Oops... The proceeding page for interim meeting seems not be as intelligent=
 as the proceeding pages of IETF meetings. Our proceeding page does not aut=
onomically show the slides. I have sent an email to IETF secretary to ask t=
hem to fix it. Meanwhile, in this
 email, here is the links for each presentations:<br>
<br>
Slide Filename Edit Replace Delete<br>
Chair Slides<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-0.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-0.pdf</a><br>
<br>
Introduction to Network Machine Learning &amp; NMLRG<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-1.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-1.pdf</a><br>
<br>
Data Collection and Analysis At High Security Lab<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-2.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-2.pdf</a><br>
<br>
Use Cases of Applying Machine Learning Mechanism with Network Traffic<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-3.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-3.pdf</a><br>
<br>
Mobile network state characterization and prediction<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-4.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-4.pdf</a><br>
<br>
Learning how to route<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-5.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-5.pdf</a><br>
<br>
Regards,<br>
<br>
Sheng<br>
________________________________________<br>
From: nmlrg [<a href=3D"mailto:nmlrg-bounces@irtf.org" target=3D"_blank">nm=
lrg-bounces@irtf.org</a>] on behalf of Sheng Jiang [<a href=3D"mailto:jiang=
sheng@huawei.com" target=3D"_blank">jiangsheng@huawei.com</a>]<br>
Sent: 28 June 2016 21:03<br>
To: <a href=3D"mailto:nmlrg@irtf.org" target=3D"_blank">nmlrg@irtf.org</a><=
br>
Subject: [Nmlrg] slides of NMLRG #3, June 27th, Athens<br>
<br>
Hi, nmlrg,<br>
<br>
All slides that have been presented in our NMLRG #3 meeting, June 27th, 201=
6, Athens, Greece, with EUCNC2016, have been uploaded. They can be accessed=
 through below link<br>
<br>
<a href=3D"https://www.ietf.org/proceedings/interim/2016/06/27/nmlrg/procee=
dings.html" rel=3D"noreferrer" target=3D"_blank">https://www.ietf.org/proce=
edings/interim/2016/06/27/nmlrg/proceedings.html</a><br>
<br>
Best regards,<br>
<br>
Sheng<br>
_______________________________________________<br>
nmlrg mailing list<br>
<a href=3D"mailto:nmlrg@irtf.org" target=3D"_blank">nmlrg@irtf.org</a><br>
<a href=3D"https://www.irtf.org/mailman/listinfo/nmlrg" rel=3D"noreferrer" =
target=3D"_blank">https://www.irtf.org/mailman/listinfo/nmlrg</a><br>
_______________________________________________<br>
nmlrg mailing list<br>
<a href=3D"mailto:nmlrg@irtf.org" target=3D"_blank">nmlrg@irtf.org</a><br>
<a href=3D"https://www.irtf.org/mailman/listinfo/nmlrg" rel=3D"noreferrer" =
target=3D"_blank">https://www.irtf.org/mailman/listinfo/nmlrg</a><br>
</blockquote>
</div>
<br>
</div>
</div>
</div></div></div>
</div>
</div>

</blockquote></div><br></div></div>

--001a1144404e5b125405366c811f--


From nobody Wed Jun 29 10:59:33 2016
Return-Path: <dmm@1-4-5.net>
X-Original-To: nmlrg@ietfa.amsl.com
Delivered-To: nmlrg@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id B72F512D548 for <nmlrg@ietfa.amsl.com>; Wed, 29 Jun 2016 10:59:32 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.599
X-Spam-Level: 
X-Spam-Status: No, score=-2.599 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, HTML_MESSAGE=0.001, RCVD_IN_DNSWL_LOW=-0.7] autolearn=ham autolearn_force=no
Authentication-Results: ietfa.amsl.com (amavisd-new); dkim=pass (2048-bit key) header.d=1-4-5-net.20150623.gappssmtp.com
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id IsaT3XDKd-3O for <nmlrg@ietfa.amsl.com>; Wed, 29 Jun 2016 10:59:31 -0700 (PDT)
Received: from mail-it0-x22a.google.com (mail-it0-x22a.google.com [IPv6:2607:f8b0:4001:c0b::22a]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by ietfa.amsl.com (Postfix) with ESMTPS id 0F9D212B01C for <nmlrg@irtf.org>; Wed, 29 Jun 2016 10:59:31 -0700 (PDT)
Received: by mail-it0-x22a.google.com with SMTP id h190so48454212ith.1 for <nmlrg@irtf.org>; Wed, 29 Jun 2016 10:59:31 -0700 (PDT)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1-4-5-net.20150623.gappssmtp.com; s=20150623; h=mime-version:from:date:message-id:subject:to; bh=J14H5RXCk/nZKP9JOXRC6kOwWzgNXXKUTE+bT25go3o=; b=bVn8hPeYiJel/Lh9yIOg9+HSP/mK6Vq2SbhKOnaGavsIN9D2GXpPFM6N190MDRUsbc 6sf7K9yVHzxnucytl5dlTbgIKstTsTYtwxBk2Yzk48Zsjim7GMIO1A2sBvrptWEV+sdj iG49AAhs7l2VL3wc4XuQPRA2tvMPVwNFyBZGjHWyFv54Z/eJbvKl8zmwNhUlJDDMciJK mErqzLJPWeQ2ZtV/YtnvI+25f1VksiOizOHw0YH1u0KOlMU3LMOuOKU/96XyvvW//eKo yfTmDID6UoVPaAmBMcyrC2hzfW48yS1qMmdoZeF+qwdwGI0Mq9ZcLj/cMmpbZTzo+Stb NI2A==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:from:date:message-id:subject:to; bh=J14H5RXCk/nZKP9JOXRC6kOwWzgNXXKUTE+bT25go3o=; b=DMgChOl9RqTf9bQgaG7YmKsaMcUgban6fOu743Ay5gLddt2ONYtOEuclIk666BEzRy Lo6CtK8HPtvaftFMcCNVbKLiJDwdSNaVNu3xhXUORpkqwe2OCebNVhRMUUXr5Gdw2Pv9 6aCIHei1WqT7Vft1TPJzmtr1G9RjD0a2vpWQiJGD70kUPSc26vLDTIiFw2UtjJYCHB5T 0ig3FTqCdETs+7St6vfcrRlWbwUlxxkvUkTYY345obKfpLYc1SXZNcJSNAMEprybcS+B waqANKVgfJlLzchLbfzXn20Q1/D8mnb2CtJwvmf39/AFSAEvr8fhyWkEOQ2Qduc9sdzs peIA==
X-Gm-Message-State: ALyK8tLVEyKcPK/dm95yV+vVQAOx84Uf360/G3Ih9P52f1pqOGLafleMnoM58KSpTB7zGlKswc8vQV1pO5Z+Zw==
X-Received: by 10.36.73.143 with SMTP id e15mr2031642itd.0.1467223170209; Wed, 29 Jun 2016 10:59:30 -0700 (PDT)
MIME-Version: 1.0
Received: by 10.64.148.133 with HTTP; Wed, 29 Jun 2016 10:59:29 -0700 (PDT)
X-Originating-IP: [128.223.156.253]
From: David Meyer <dmm@1-4-5.net>
Date: Wed, 29 Jun 2016 10:59:29 -0700
Message-ID: <CAHiKxWihP=9jQDb2X1DB3EEoNPoB+WSnmRJAU9zo-pTf9tjq=w@mail.gmail.com>
To: nmlrg@irtf.org
Content-Type: multipart/alternative; boundary=001a11c14d24de305d05366e835f
Archived-At: <https://mailarchive.ietf.org/arch/msg/nmlrg/g-dE1z4Y6AbIJsppMlh4646OGkg>
Subject: [Nmlrg] One comment on the proposed charter...
X-BeenThere: nmlrg@irtf.org
X-Mailman-Version: 2.1.17
Precedence: list
List-Id: Network Machine Learning Research Group <nmlrg.irtf.org>
List-Unsubscribe: <https://www.irtf.org/mailman/options/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=unsubscribe>
List-Archive: <https://mailarchive.ietf.org/arch/browse/nmlrg/>
List-Post: <mailto:nmlrg@irtf.org>
List-Help: <mailto:nmlrg-request@irtf.org?subject=help>
List-Subscribe: <https://www.irtf.org/mailman/listinfo/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=subscribe>
X-List-Received-Date: Wed, 29 Jun 2016 17:59:32 -0000

--001a11c14d24de305d05366e835f
Content-Type: text/plain; charset=UTF-8

All,

I was just reviewing the proposed charter and one thing we might consider
adding is the open database item (formally standardized datasets). We need
this not only to make progress and to prove out how various approaches work
(or don't work), whether new approaches improve on existing methods, and
the like.

More generally, the machine learning community has a long standing
tradition of open source, open data, and even open models (AlexNet,
GoogLeNet, etc are excellent examples of the later). The NMLRG should
strive to carry on these traditions, and they should probably be reflected
in the charter. Perhaps something like:

To "The Network Machine Learning Research Group (NMLRG) provides a forum
for researchers to explore the potential of machine learning technologies
for networks. In particular, the NMLRG will work on potential approaches
that apply machine learning technologies in network control, network
management, and supplying network data for upper-layer applications.'

add something like


"In addition and to the extent possible, approaches proposed in the NMLRG
should be documented not only with traditional written documentation but
also with an open source reference implementation and open access to the
data used to verify the approach."

or similar, again with the goal of supporting/carrying on the open science
approach that has been so successful to date in the ML community.

Thanks,

--dmm

--001a11c14d24de305d05366e835f
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div><br></div><div>All,</div><div><br></div><div>I was ju=
st reviewing the proposed charter and one thing we might consider adding is=
 the open database item (formally standardized datasets). We need this not =
only to make progress and to prove out how various approaches work (or don&=
#39;t work), whether new approaches improve on existing methods, and the li=
ke.</div><div><br></div><div>More generally, the machine learning community=
 has a long standing tradition of open source, open data, and even open mod=
els (AlexNet, GoogLeNet, etc are excellent examples of the later). The NMLR=
G should strive to carry on these traditions, and they should probably be r=
eflected in the charter. Perhaps something like:</div><div><br></div><div>T=
o &quot;The Network Machine Learning Research Group (NMLRG) provides a foru=
m for researchers to explore the potential of machine learning technologies=
 for networks. In particular, the NMLRG will work on potential approaches t=
hat apply machine learning technologies in network control, network managem=
ent, and supplying network data for upper-layer applications.&#39; =C2=A0</=
div><div><br></div><div>add something like</div><div><br></div><div><br></d=
iv><div>&quot;In addition and to the extent possible, approaches proposed i=
n the NMLRG should be documented not only with traditional written document=
ation but also with an open source reference implementation and open access=
 to the data used to verify the approach.&quot;</div><div><br></div><div>or=
 similar, again with the goal of supporting/carrying on the open science ap=
proach that has been so successful to date in the ML community.</div><div><=
br></div><div>Thanks,</div><div><br></div><div>--dmm</div><div><br></div></=
div>

--001a11c14d24de305d05366e835f--


From nobody Wed Jun 29 11:09:36 2016
Return-Path: <dmm@1-4-5.net>
X-Original-To: nmlrg@ietfa.amsl.com
Delivered-To: nmlrg@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id A16A712D522 for <nmlrg@ietfa.amsl.com>; Wed, 29 Jun 2016 11:09:34 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.599
X-Spam-Level: 
X-Spam-Status: No, score=-2.599 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, HTML_MESSAGE=0.001, RCVD_IN_DNSWL_LOW=-0.7] autolearn=ham autolearn_force=no
Authentication-Results: ietfa.amsl.com (amavisd-new); dkim=pass (2048-bit key) header.d=1-4-5-net.20150623.gappssmtp.com
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id gQ95Fkdl6rKM for <nmlrg@ietfa.amsl.com>; Wed, 29 Jun 2016 11:09:32 -0700 (PDT)
Received: from mail-it0-x22d.google.com (mail-it0-x22d.google.com [IPv6:2607:f8b0:4001:c0b::22d]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by ietfa.amsl.com (Postfix) with ESMTPS id 343BD12D1D8 for <nmlrg@irtf.org>; Wed, 29 Jun 2016 11:09:32 -0700 (PDT)
Received: by mail-it0-x22d.google.com with SMTP id a5so126571897ita.1 for <nmlrg@irtf.org>; Wed, 29 Jun 2016 11:09:32 -0700 (PDT)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1-4-5-net.20150623.gappssmtp.com; s=20150623; h=mime-version:in-reply-to:references:from:date:message-id:subject:to :cc; bh=TT367YYiCfaCqhdwAtGz69jH5sTKNfu/1Pu5ouk7L9g=; b=0a3zhEJin0yE7G0RBRwooNbfYPui223/GlQh7dC8KYqeIzQbbpL14ps6zrNDBc0nzY LWHRZx9PTRgBJRUg+uErmecxHjH4ughDW1vTLkdRzLtxoktljdIDJKBeh9hPIxhiyyAr i1EdBtY2ewanVROJ06aQIpB3/WlueOzHRXNH3PF0thoEh1K4D950ILHSiTb/S88wRlCv JrN3NF/w7ivyKZPKAvCl9cLWsa8GGC7g4e8MLp9hfYhVPqk1kyOo447mxUn4O0v/9SSv 64PIGntlt324yVANW1YSqpvEHFsT1VCfoJZ4NECj538NQbZRXyiyRfTWOGO9ZrA1DaKX 30iw==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:from:date :message-id:subject:to:cc; bh=TT367YYiCfaCqhdwAtGz69jH5sTKNfu/1Pu5ouk7L9g=; b=DNOUeUo7FLWjasr+ksDfRCCfa1UdNVnAMSRk35ChwfdxmyY+Wa+/XmgbU9VFxvkzXK PaDReMtvCOMwvOfylyKkiTeUMLmq584qPKup7IabndF/3jSzj0nfB6E9cYvR1Nsbo+Cl p8rYb64bfSCg5zMzExNhmEIImMNu4Ce1hoKg0Y+nANKgH7dXx6moZdckGV5HWRENWb+B SkES2qzzI9dT9R3j/eKeUWtafsbo9McKD6XnaX1Jbs2EE9hzuMfcqkxsvW++TIRG9STu URROneR6sfqzrihneIlHvqSLNnlRtvr+xJL2mA+WEOVLIJTtlTSGGn7aPbpf60hb8kDl 0ndw==
X-Gm-Message-State: ALyK8tJFy5/z0t8kkR0S1JQkNWm2HAKcHhcwoTaRlSAOkVQKB4dSq7V8pwHvlxGH1ayw4s7VRVdltmF0Ao8qTw==
X-Received: by 10.36.101.74 with SMTP id u71mr11429190itb.92.1467223771449; Wed, 29 Jun 2016 11:09:31 -0700 (PDT)
MIME-Version: 1.0
Received: by 10.64.148.133 with HTTP; Wed, 29 Jun 2016 11:09:30 -0700 (PDT)
X-Originating-IP: [128.223.156.253]
In-Reply-To: <5D36713D8A4E7348A7E10DF7437A4B927CA893E9@NKGEML515-MBX.china.huawei.com>
References: <5D36713D8A4E7348A7E10DF7437A4B927CA893E9@NKGEML515-MBX.china.huawei.com>
From: David Meyer <dmm@1-4-5.net>
Date: Wed, 29 Jun 2016 11:09:30 -0700
Message-ID: <CAHiKxWi_mru0CG1ysoSVvN+beadHnmat2hzo5qeAQFudeYv75w@mail.gmail.com>
To: Sheng Jiang <jiangsheng@huawei.com>
Content-Type: multipart/alternative; boundary=001a1145bcf2b4236905366ea747
Archived-At: <https://mailarchive.ietf.org/arch/msg/nmlrg/wsPbSTtTk-B3pD4FhTjIGBU2ar0>
Cc: "nmlrg@irtf.org" <nmlrg@irtf.org>
Subject: Re: [Nmlrg] links to each slide of presentations//RE: slides of NMLRG #3, June 27th, Athens
X-BeenThere: nmlrg@irtf.org
X-Mailman-Version: 2.1.17
Precedence: list
List-Id: Network Machine Learning Research Group <nmlrg.irtf.org>
List-Unsubscribe: <https://www.irtf.org/mailman/options/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=unsubscribe>
List-Archive: <https://mailarchive.ietf.org/arch/browse/nmlrg/>
List-Post: <mailto:nmlrg@irtf.org>
List-Help: <mailto:nmlrg-request@irtf.org?subject=help>
List-Subscribe: <https://www.irtf.org/mailman/listinfo/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=subscribe>
X-List-Received-Date: Wed, 29 Jun 2016 18:09:34 -0000

--001a1145bcf2b4236905366ea747
Content-Type: text/plain; charset=UTF-8

This question is on this use case:

Mobile network state characterization and prediction
https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-4.pdf

Isn't it the case that you will know if something is registered to the MBB,
MMC, or MCC slice at device authentication time? If not what is the
scenario in which you don't know?

Thanks,

--dmm



On Tue, Jun 28, 2016 at 7:03 AM, Sheng Jiang <jiangsheng@huawei.com> wrote:

> Oops... The proceeding page for interim meeting seems not be as
> intelligent as the proceeding pages of IETF meetings. Our proceeding page
> does not autonomically show the slides. I have sent an email to IETF
> secretary to ask them to fix it. Meanwhile, in this email, here is the
> links for each presentations:
>
> Slide Filename Edit Replace Delete
> Chair Slides
>
> https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-0.pdf
>
> Introduction to Network Machine Learning & NMLRG
>
> https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-1.pdf
>
> Data Collection and Analysis At High Security Lab
>
> https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-2.pdf
>
> Use Cases of Applying Machine Learning Mechanism with Network Traffic
>
> https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-3.pdf
>
> Mobile network state characterization and prediction
>
> https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-4.pdf
>
> Learning how to route
>
> https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-5.pdf
>
> Regards,
>
> Sheng
> ________________________________________
> From: nmlrg [nmlrg-bounces@irtf.org] on behalf of Sheng Jiang [
> jiangsheng@huawei.com]
> Sent: 28 June 2016 21:03
> To: nmlrg@irtf.org
> Subject: [Nmlrg] slides of NMLRG #3, June 27th, Athens
>
> Hi, nmlrg,
>
> All slides that have been presented in our NMLRG #3 meeting, June 27th,
> 2016, Athens, Greece, with EUCNC2016, have been uploaded. They can be
> accessed through below link
>
> https://www.ietf.org/proceedings/interim/2016/06/27/nmlrg/proceedings.html
>
> Best regards,
>
> Sheng
> _______________________________________________
> nmlrg mailing list
> nmlrg@irtf.org
> https://www.irtf.org/mailman/listinfo/nmlrg
> _______________________________________________
> nmlrg mailing list
> nmlrg@irtf.org
> https://www.irtf.org/mailman/listinfo/nmlrg
>

--001a1145bcf2b4236905366ea747
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr">This question is on this use case:<div><br><div><div>Mobil=
e network state characterization and prediction</div><div><a href=3D"https:=
//www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016=
-nmlrg-01-4.pdf">https://www.ietf.org/proceedings/interim-2016-nmlrg-01/sli=
des/slides-interim-2016-nmlrg-01-4.pdf</a></div></div><div><br></div><div>I=
sn&#39;t it the case that you will know if something is registered to the M=
BB, MMC, or MCC slice at device authentication time? If not what is the sce=
nario in which you don&#39;t know?</div><div><br></div><div>Thanks,</div><d=
iv><br></div><div>--dmm</div><div><br></div><div><br><div class=3D"gmail_ex=
tra"><br><div class=3D"gmail_quote">On Tue, Jun 28, 2016 at 7:03 AM, Sheng =
Jiang <span dir=3D"ltr">&lt;<a href=3D"mailto:jiangsheng@huawei.com" target=
=3D"_blank">jiangsheng@huawei.com</a>&gt;</span> wrote:<br><blockquote clas=
s=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-left-width:1px;b=
order-left-style:solid;border-left-color:rgb(204,204,204);padding-left:1ex"=
>Oops... The proceeding page for interim meeting seems not be as intelligen=
t as the proceeding pages of IETF meetings. Our proceeding page does not au=
tonomically show the slides. I have sent an email to IETF secretary to ask =
them to fix it. Meanwhile, in this email, here is the links for each presen=
tations:<br>
<br>
Slide Filename Edit Replace Delete<br>
Chair Slides<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-0.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-0.pdf</a><br>
<br>
Introduction to Network Machine Learning &amp; NMLRG<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-1.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-1.pdf</a><br>
<br>
Data Collection and Analysis At High Security Lab<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-2.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-2.pdf</a><br>
<br>
Use Cases of Applying Machine Learning Mechanism with Network Traffic<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-3.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-3.pdf</a><br>
<br>
Mobile network state characterization and prediction<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-4.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-4.pdf</a><br>
<br>
Learning how to route<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-5.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-5.pdf</a><br>
<br>
Regards,<br>
<br>
Sheng<br>
________________________________________<br>
From: nmlrg [<a href=3D"mailto:nmlrg-bounces@irtf.org">nmlrg-bounces@irtf.o=
rg</a>] on behalf of Sheng Jiang [<a href=3D"mailto:jiangsheng@huawei.com">=
jiangsheng@huawei.com</a>]<br>
Sent: 28 June 2016 21:03<br>
To: <a href=3D"mailto:nmlrg@irtf.org">nmlrg@irtf.org</a><br>
Subject: [Nmlrg] slides of NMLRG #3, June 27th, Athens<br>
<br>
Hi, nmlrg,<br>
<br>
All slides that have been presented in our NMLRG #3 meeting, June 27th, 201=
6, Athens, Greece, with EUCNC2016, have been uploaded. They can be accessed=
 through below link<br>
<br>
<a href=3D"https://www.ietf.org/proceedings/interim/2016/06/27/nmlrg/procee=
dings.html" rel=3D"noreferrer" target=3D"_blank">https://www.ietf.org/proce=
edings/interim/2016/06/27/nmlrg/proceedings.html</a><br>
<br>
Best regards,<br>
<br>
Sheng<br>
_______________________________________________<br>
nmlrg mailing list<br>
<a href=3D"mailto:nmlrg@irtf.org">nmlrg@irtf.org</a><br>
<a href=3D"https://www.irtf.org/mailman/listinfo/nmlrg" rel=3D"noreferrer" =
target=3D"_blank">https://www.irtf.org/mailman/listinfo/nmlrg</a><br>
_______________________________________________<br>
nmlrg mailing list<br>
<a href=3D"mailto:nmlrg@irtf.org">nmlrg@irtf.org</a><br>
<a href=3D"https://www.irtf.org/mailman/listinfo/nmlrg" rel=3D"noreferrer" =
target=3D"_blank">https://www.irtf.org/mailman/listinfo/nmlrg</a><br>
</blockquote></div><br></div></div></div></div>

--001a1145bcf2b4236905366ea747--


From nobody Wed Jun 29 11:48:28 2016
Return-Path: <dmm@1-4-5.net>
X-Original-To: nmlrg@ietfa.amsl.com
Delivered-To: nmlrg@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 48DB712D0DC for <nmlrg@ietfa.amsl.com>; Wed, 29 Jun 2016 11:48:27 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.599
X-Spam-Level: 
X-Spam-Status: No, score=-2.599 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, HTML_MESSAGE=0.001, RCVD_IN_DNSWL_LOW=-0.7] autolearn=ham autolearn_force=no
Authentication-Results: ietfa.amsl.com (amavisd-new); dkim=pass (2048-bit key) header.d=1-4-5-net.20150623.gappssmtp.com
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id D02v4B3x3fwb for <nmlrg@ietfa.amsl.com>; Wed, 29 Jun 2016 11:48:25 -0700 (PDT)
Received: from mail-it0-x243.google.com (mail-it0-x243.google.com [IPv6:2607:f8b0:4001:c0b::243]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by ietfa.amsl.com (Postfix) with ESMTPS id B117912B05C for <nmlrg@irtf.org>; Wed, 29 Jun 2016 11:48:25 -0700 (PDT)
Received: by mail-it0-x243.google.com with SMTP id h190so6274681ith.3 for <nmlrg@irtf.org>; Wed, 29 Jun 2016 11:48:25 -0700 (PDT)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1-4-5-net.20150623.gappssmtp.com; s=20150623; h=mime-version:from:date:message-id:subject:to; bh=+fs8KGjwTEgPrkNDzoQLh6YWNECMAv8jCz4CYsFVVf8=; b=x2Fn8JzHOsmOKftgbWDgLKDuBqDFCQO6I9Pzy7VnLLXAl1JOUEfQFli6O9Yr4LSOP9 zJpp+bBzaiK1I2LiE13dmpSFjwBJ5joNs8jrCGrVb8GvMjJu8aVUqL+WVXfH46zTCILv j58HS+r2HIOM/Dv78KbOcAD6ZJ0LhnLcKUU3e9LNa3/eI+pBik/SVLSIZjqrOBbtS2+T 7NEvMQ2EMxyWEqoZolIbf9EYRdXAhf5w/D2bNGhajgT/Prbc3TpZQyPtmCARRljNXdFl BjrI9se4N9kpoj92Rv6aRloy7ycXJ/as9B9WnKH+WbijZmFttDyMAvFxuFxrXA5URs7i Fi1A==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:from:date:message-id:subject:to; bh=+fs8KGjwTEgPrkNDzoQLh6YWNECMAv8jCz4CYsFVVf8=; b=fPPmTKBfGCNb0j/H2XSIQYckmE8skLEv7sgIPtFiRUIbZX6nszx1M5E/8INLb/3UEg 2hnmfhYG6n1LRdlHJ+0GvO2NVko56je8HSb0MTn/BUwnuszHzx40l2Xo51zhPtt01KDV ccuvAlDvlcSVD0atpnY7jMR5uwhURTxCOb8cEF8p5gBmBY1BIqk2TRVm2BbHGsK1Y1E0 PNHyuMQoAqdMTWyML4rBy2d4psl+qFVz6K4VA5N+lOheNZi5E5Hnp/CVFHGxmPzHfQ9U HYdDoCp9NjyrDPHEcv8NOOW5VNKViDNuTvtXkWIse0bTLgqFmHglcSnZ/kJIjrB8ewh7 Q/UQ==
X-Gm-Message-State: ALyK8tK8s7YJLcq9jS3At97924p0AteasJhzp3/KCaQ53VV+BXH8gMEEp7TmmA07+ak2s0uboJjLiOaTQCFYLg==
X-Received: by 10.36.89.83 with SMTP id p80mr22260208itb.7.1467226104766; Wed, 29 Jun 2016 11:48:24 -0700 (PDT)
MIME-Version: 1.0
Received: by 10.64.148.133 with HTTP; Wed, 29 Jun 2016 11:48:24 -0700 (PDT)
X-Originating-IP: [128.223.156.253]
From: David Meyer <dmm@1-4-5.net>
Date: Wed, 29 Jun 2016 11:48:24 -0700
Message-ID: <CAHiKxWiqk8XAp-XAhmnp3K9vmNOW+YHJ_pjCpFdE_86NdoepYQ@mail.gmail.com>
To: nmlrg@irtf.org
Content-Type: multipart/alternative; boundary=001a11434ae4c7ba3f05366f3280
Archived-At: <https://mailarchive.ietf.org/arch/msg/nmlrg/prRLDpD0FIkA_0AdRtL6Zy_mWeE>
Subject: [Nmlrg] new ways of dealing with sparse categorical data like IP addresses (Google Wide & Deep)
X-BeenThere: nmlrg@irtf.org
X-Mailman-Version: 2.1.17
Precedence: list
List-Id: Network Machine Learning Research Group <nmlrg.irtf.org>
List-Unsubscribe: <https://www.irtf.org/mailman/options/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=unsubscribe>
List-Archive: <https://mailarchive.ietf.org/arch/browse/nmlrg/>
List-Post: <mailto:nmlrg@irtf.org>
List-Help: <mailto:nmlrg-request@irtf.org?subject=help>
List-Subscribe: <https://www.irtf.org/mailman/listinfo/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=subscribe>
X-List-Received-Date: Wed, 29 Jun 2016 18:48:27 -0000

--001a11434ae4c7ba3f05366f3280
Content-Type: text/plain; charset=UTF-8

Check out Wide & Deep:
https://research.googleblog.com/2016/06/wide-deep-learning-better-together-with.html

Among other things, Wide & Deep is

"useful for generic large-scale regression and classification problems with
sparse inputs (categorical features with a large number of possible feature
values), such as recommender systems, search, and ranking problems."

In the network space we have many such sparse categorical features such as
IPv{4,6} addresses. Wide & Deep might give us a few  hints about how we
might approach this problem.

--dmm

--001a11434ae4c7ba3f05366f3280
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr">Check out Wide &amp; Deep:=C2=A0<a href=3D"https://researc=
h.googleblog.com/2016/06/wide-deep-learning-better-together-with.html">http=
s://research.googleblog.com/2016/06/wide-deep-learning-better-together-with=
.html</a><div><br></div><div><div>Among other things, Wide &amp; Deep is</d=
iv><div><br></div><div>&quot;useful for generic large-scale regression and =
classification problems with sparse inputs (categorical features with a lar=
ge number of possible feature values), such as recommender systems, search,=
 and ranking problems.&quot;</div></div><div><br></div><div>In the network =
space we have many such sparse categorical features such as IPv{4,6} addres=
ses. Wide &amp; Deep might give us a few =C2=A0hints about how we might app=
roach this problem.</div><div><br></div><div>--dmm</div><div><br></div><div=
><br></div></div>

--001a11434ae4c7ba3f05366f3280--


From nobody Wed Jun 29 13:11:42 2016
Return-Path: <networkn3rd@gmail.com>
X-Original-To: nmlrg@ietfa.amsl.com
Delivered-To: nmlrg@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id C1FCD12D6A2 for <nmlrg@ietfa.amsl.com>; Wed, 29 Jun 2016 13:11:40 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.699
X-Spam-Level: 
X-Spam-Status: No, score=-2.699 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FREEMAIL_FROM=0.001, HTML_MESSAGE=0.001, RCVD_IN_DNSWL_LOW=-0.7, SPF_PASS=-0.001] autolearn=ham autolearn_force=no
Authentication-Results: ietfa.amsl.com (amavisd-new); dkim=pass (2048-bit key) header.d=gmail.com
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id O5WxqsxDPhvb for <nmlrg@ietfa.amsl.com>; Wed, 29 Jun 2016 13:11:38 -0700 (PDT)
Received: from mail-ob0-x22c.google.com (mail-ob0-x22c.google.com [IPv6:2607:f8b0:4003:c01::22c]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by ietfa.amsl.com (Postfix) with ESMTPS id DC4CD12D6A7 for <nmlrg@irtf.org>; Wed, 29 Jun 2016 13:11:30 -0700 (PDT)
Received: by mail-ob0-x22c.google.com with SMTP id ru5so41447330obc.1 for <nmlrg@irtf.org>; Wed, 29 Jun 2016 13:11:30 -0700 (PDT)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20120113;  h=mime-version:in-reply-to:references:from:date:message-id:subject:to :cc; bh=4dgE7S2WEjLHjLtFzTrudcRS6mXu1uBTAPdoJOMBEVQ=; b=YXmMH8u5BJEgcFAk9SMucbB23jHX1iSruqosoUQwwiyjG6aPwLiG9/lL/uImh6HNYw /hHlHcNhXdvS+VccZp0OsjkmWwdUuA1rJuc4U1hUoBnBLfnHk6RvtzDbPyfEMrFs5BuC N4r1PdKzxVIDNPwfnydMkNEhhfCyUwxQIkN2bnRPLdmuV0gS3izex9ggo3mvFySowCkt y9NAdzdiA9pkHXzIQk3ISiPgQDLmo/otUORZxtKZ8ASlf24pX4kaOCsYkN5J05wY5Jyv qZdQT6CGw0fGB909FaEQvUp+XPLktD4dZ11t1hVqlpgvHNEgkjdpRZEOT9VLqNOjC+3B +FgQ==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:from:date :message-id:subject:to:cc; bh=4dgE7S2WEjLHjLtFzTrudcRS6mXu1uBTAPdoJOMBEVQ=; b=HxAyLpvsPLMcaDmD85Qx16LRiHk9D5H7VLIiHDE9Vo5oARK4xzT/4xPwB4su7WCAgY Me3n6W8kjXCGSDP5gen8cqBzAyKh7ayba1zU5KqvkO0Hh0AkK4cCuKGAJ5LMVLzApbkD FXpea+ys8hi9RiEco6psuk62BCt42Y36hQJF9t6gQWPd6ajjS5gIJA+72tJ/xPV6zck2 w0Q2rT5OMVJCngIAX/jn3pR0nHTnqHlZht6okF31VF/ZOl0q1FuVEFmgADXjkWcUceD3 aejn7bFYoKsyb8r73ZHb0dklY69VTsXWYdLBZwq5g6KQxcZM1AaKrYoKo1f+IaCoCQq9 5k9w==
X-Gm-Message-State: ALyK8tKT/KqJr7a6JTtgFUXmx4zfdDdKH16nME6g6FaY/WnNrU0MQDcQsPI23I/G/aW5jWFhsJv8r3CH/dGGog==
X-Received: by 10.157.40.111 with SMTP id h44mr7623613otd.104.1467231090225; Wed, 29 Jun 2016 13:11:30 -0700 (PDT)
MIME-Version: 1.0
Received: by 10.202.71.199 with HTTP; Wed, 29 Jun 2016 13:11:29 -0700 (PDT)
In-Reply-To: <CAHiKxWihP=9jQDb2X1DB3EEoNPoB+WSnmRJAU9zo-pTf9tjq=w@mail.gmail.com>
References: <CAHiKxWihP=9jQDb2X1DB3EEoNPoB+WSnmRJAU9zo-pTf9tjq=w@mail.gmail.com>
From: Ed Henry <networkn3rd@gmail.com>
Date: Wed, 29 Jun 2016 16:11:29 -0400
Message-ID: <CADnDDRFrMp684SMLiGjNE_hszZ8CABG5o2tGdZp9U6_LOJYetw@mail.gmail.com>
To: David Meyer <dmm@1-4-5.net>
Content-Type: multipart/alternative; boundary=001a11415770efc43b0536705bba
Archived-At: <https://mailarchive.ietf.org/arch/msg/nmlrg/0VaN1JXoScgZlkkNv8IUIuv8MN0>
Cc: nmlrg@irtf.org
Subject: Re: [Nmlrg] One comment on the proposed charter...
X-BeenThere: nmlrg@irtf.org
X-Mailman-Version: 2.1.17
Precedence: list
List-Id: Network Machine Learning Research Group <nmlrg.irtf.org>
List-Unsubscribe: <https://www.irtf.org/mailman/options/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=unsubscribe>
List-Archive: <https://mailarchive.ietf.org/arch/browse/nmlrg/>
List-Post: <mailto:nmlrg@irtf.org>
List-Help: <mailto:nmlrg-request@irtf.org?subject=help>
List-Subscribe: <https://www.irtf.org/mailman/listinfo/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=subscribe>
X-List-Received-Date: Wed, 29 Jun 2016 20:11:41 -0000

--001a11415770efc43b0536705bba
Content-Type: text/plain; charset=UTF-8

On Wed, Jun 29, 2016 at 1:59 PM, David Meyer <dmm@1-4-5.net> wrote:

>
> All,
>
> I was just reviewing the proposed charter and one thing we might consider
> adding is the open database item (formally standardized datasets). We need
> this not only to make progress and to prove out how various approaches work
> (or don't work), whether new approaches improve on existing methods, and
> the like.
>
> More generally, the machine learning community has a long standing
> tradition of open source, open data, and even open models (AlexNet,
> GoogLeNet, etc are excellent examples of the later). The NMLRG should
> strive to carry on these traditions, and they should probably be reflected
> in the charter. Perhaps something like:
>

+1 On keeping this long standing tradition.


> To "The Network Machine Learning Research Group (NMLRG) provides a forum
> for researchers to explore the potential of machine learning technologies
> for networks. In particular, the NMLRG will work on potential approaches
> that apply machine learning technologies in network control, network
> management, and supplying network data for upper-layer applications.'
>
> add something like
>
>
> "In addition and to the extent possible, approaches proposed in the NMLRG
> should be documented not only with traditional written documentation but
> also with an open source reference implementation and open access to the
> data used to verify the approach."
>

> or similar, again with the goal of supporting/carrying on the open science
> approach that has been so successful to date in the ML community.
>

+1 again. Would it be possible to use a hosted service such as Github or
the like for hosting code, documentation, and data-set(s)? I think,
traditionally, there was some adversity to using a hosted service like
Github, at least within the IETF, but in general the ML community rallies
around this platform. So it would almost make sense in that respect.


>
> Thanks,
>
> --dmm
>
>
> _______________________________________________
> nmlrg mailing list
> nmlrg@irtf.org
> https://www.irtf.org/mailman/listinfo/nmlrg
>
>

--001a11415770efc43b0536705bba
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><br><div class=3D"gmail_extra"><br><div class=3D"gmail_quo=
te">On Wed, Jun 29, 2016 at 1:59 PM, David Meyer <span dir=3D"ltr">&lt;<a h=
ref=3D"mailto:dmm@1-4-5.net" target=3D"_blank">dmm@1-4-5.net</a>&gt;</span>=
 wrote:<br><blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.=
8ex;border-left-width:1px;border-left-style:solid;border-left-color:rgb(204=
,204,204);padding-left:1ex"><div dir=3D"ltr"><div><br></div><div>All,</div>=
<div><br></div><div>I was just reviewing the proposed charter and one thing=
 we might consider adding is the open database item (formally standardized =
datasets). We need this not only to make progress and to prove out how vari=
ous approaches work (or don&#39;t work), whether new approaches improve on =
existing methods, and the like.</div><div><br></div><div>More generally, th=
e machine learning community has a long standing tradition of open source, =
open data, and even open models (AlexNet, GoogLeNet, etc are excellent exam=
ples of the later). The NMLRG should strive to carry on these traditions, a=
nd they should probably be reflected in the charter. Perhaps something like=
:</div></div></blockquote><div><br></div><div>+1 On keeping this long stand=
ing tradition.</div><div><br></div><blockquote class=3D"gmail_quote" style=
=3D"margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-style:solid;=
border-left-color:rgb(204,204,204);padding-left:1ex"><div dir=3D"ltr"><div>=
<br></div><div>To &quot;The Network Machine Learning Research Group (NMLRG)=
 provides a forum for researchers to explore the potential of machine learn=
ing technologies for networks. In particular, the NMLRG will work on potent=
ial approaches that apply machine learning technologies in network control,=
 network management, and supplying network data for upper-layer application=
s.&#39; =C2=A0</div><div><br></div><div>add something like</div><div><br></=
div><div><br></div><div>&quot;In addition and to the extent possible, appro=
aches proposed in the NMLRG should be documented not only with traditional =
written documentation but also with an open source reference implementation=
 and open access to the data used to verify the approach.&quot;</div></div>=
</blockquote><blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px =
0.8ex;border-left-width:1px;border-left-style:solid;border-left-color:rgb(2=
04,204,204);padding-left:1ex"><div dir=3D"ltr"><div><br></div><div>or simil=
ar, again with the goal of supporting/carrying on the open science approach=
 that has been so successful to date in the ML community.</div></div></bloc=
kquote><div><br></div><div>+1 again. Would it be possible to use a hosted s=
ervice such as Github or the like for hosting code, documentation, and data=
-set(s)? I think, traditionally, there was some adversity to using a hosted=
 service like Github, at least within the IETF, but in general the ML commu=
nity rallies around this platform. So it would almost make sense in that re=
spect.</div><div>=C2=A0</div><blockquote class=3D"gmail_quote" style=3D"mar=
gin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-style:solid;border-=
left-color:rgb(204,204,204);padding-left:1ex"><div dir=3D"ltr"><div><br></d=
iv><div>Thanks,</div><div><br></div><div>--dmm</div><div><br></div></div>
<br>_______________________________________________<br>
nmlrg mailing list<br>
<a href=3D"mailto:nmlrg@irtf.org">nmlrg@irtf.org</a><br>
<a href=3D"https://www.irtf.org/mailman/listinfo/nmlrg" rel=3D"noreferrer" =
target=3D"_blank">https://www.irtf.org/mailman/listinfo/nmlrg</a><br>
<br></blockquote></div><br></div></div>

--001a11415770efc43b0536705bba--


From nobody Thu Jun 30 10:35:13 2016
Return-Path: <jiangsheng@huawei.com>
X-Original-To: nmlrg@ietfa.amsl.com
Delivered-To: nmlrg@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 785EB12D0FD for <nmlrg@ietfa.amsl.com>; Thu, 30 Jun 2016 10:35:12 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -5.646
X-Spam-Level: 
X-Spam-Status: No, score=-5.646 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, HTML_MESSAGE=0.001, RCVD_IN_DNSWL_MED=-2.3, RCVD_IN_MSPIKE_H3=-0.01, RCVD_IN_MSPIKE_WL=-0.01, RP_MATCHES_RCVD=-1.426, SPF_PASS=-0.001] autolearn=ham autolearn_force=no
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id BBejzgrg_xDJ for <nmlrg@ietfa.amsl.com>; Thu, 30 Jun 2016 10:35:09 -0700 (PDT)
Received: from szxga03-in.huawei.com (szxga03-in.huawei.com [119.145.14.66]) (using TLSv1 with cipher RC4-SHA (128/128 bits)) (No client certificate requested) by ietfa.amsl.com (Postfix) with ESMTPS id 3B3FF12B00B for <nmlrg@irtf.org>; Thu, 30 Jun 2016 10:35:01 -0700 (PDT)
Received: from 172.24.1.60 (EHLO NKGEML413-HUB.china.huawei.com) ([172.24.1.60]) by szxrg03-dlp.huawei.com (MOS 4.4.3-GA FastPath queued) with ESMTP id CED85494; Fri, 01 Jul 2016 01:34:50 +0800 (CST)
Received: from NKGEML515-MBX.china.huawei.com ([fe80::a54a:89d2:c471:ff]) by NKGEML413-HUB.china.huawei.com ([10.98.56.74]) with mapi id 14.03.0235.001; Fri, 1 Jul 2016 01:34:44 +0800
From: Sheng Jiang <jiangsheng@huawei.com>
To: Ed Henry <networkn3rd@gmail.com>, David Meyer <dmm@1-4-5.net>
Thread-Topic: [Nmlrg] One comment on the proposed charter...
Thread-Index: AQHR0jAG3jOkuYLUh0WK3iovXQDfcKAAWluAgAEj+kw=
Date: Thu, 30 Jun 2016 17:34:43 +0000
Message-ID: <5D36713D8A4E7348A7E10DF7437A4B927CA89EC7@NKGEML515-MBX.china.huawei.com>
References: <CAHiKxWihP=9jQDb2X1DB3EEoNPoB+WSnmRJAU9zo-pTf9tjq=w@mail.gmail.com>,  <CADnDDRFrMp684SMLiGjNE_hszZ8CABG5o2tGdZp9U6_LOJYetw@mail.gmail.com>
In-Reply-To: <CADnDDRFrMp684SMLiGjNE_hszZ8CABG5o2tGdZp9U6_LOJYetw@mail.gmail.com>
Accept-Language: en-GB, zh-CN, en-US
Content-Language: en-GB
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
x-originating-ip: [10.47.64.200]
Content-Type: multipart/alternative; boundary="_000_5D36713D8A4E7348A7E10DF7437A4B927CA89EC7NKGEML515MBXchi_"
MIME-Version: 1.0
X-CFilter-Loop: Reflected
X-Mirapoint-Virus-RAPID-Raw: score=unknown(0), refid=str=0001.0A090204.5775583B.005B, ss=1, re=0.000, recu=0.000, reip=0.000,  cl=1, cld=1, fgs=0, ip=0.0.0.0, so=2013-05-26 15:14:31, dmn=2013-03-21 17:37:32
X-Mirapoint-Loop-Id: a0f5c8c983ae23b4f7926d17be5acd2b
Archived-At: <https://mailarchive.ietf.org/arch/msg/nmlrg/lX_2UKZX29pc_1Q0VhHrYLD04hE>
Cc: "nmlrg@irtf.org" <nmlrg@irtf.org>
Subject: Re: [Nmlrg] One comment on the proposed charter...
X-BeenThere: nmlrg@irtf.org
X-Mailman-Version: 2.1.17
Precedence: list
List-Id: Network Machine Learning Research Group <nmlrg.irtf.org>
List-Unsubscribe: <https://www.irtf.org/mailman/options/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=unsubscribe>
List-Archive: <https://mailarchive.ietf.org/arch/browse/nmlrg/>
List-Post: <mailto:nmlrg@irtf.org>
List-Help: <mailto:nmlrg-request@irtf.org?subject=help>
List-Subscribe: <https://www.irtf.org/mailman/listinfo/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=subscribe>
X-List-Received-Date: Thu, 30 Jun 2016 17:35:12 -0000

--_000_5D36713D8A4E7348A7E10DF7437A4B927CA89EC7NKGEML515MBXchi_
Content-Type: text/plain; charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable

Thanks David and Ed,



This is indeed a good suggestion. I also like the openness although our tra=
ditional players in the network area, like operators and vendors, may not b=
e as open as ML communities. So, I may soften requirements a little bit.  W=
e could certainly add such text when the next time we modify the charter. T=
here may be a chance to do so sometime later this year when we formally app=
roved as a long-running IRTF research group, if we could get approval.



Regards,



Sheng



________________________________
From: nmlrg [nmlrg-bounces@irtf.org] on behalf of Ed Henry [networkn3rd@gma=
il.com]
Sent: 30 June 2016 4:11
To: David Meyer
Cc: nmlrg@irtf.org
Subject: Re: [Nmlrg] One comment on the proposed charter...



On Wed, Jun 29, 2016 at 1:59 PM, David Meyer <dmm@1-4-5.net<mailto:dmm@1-4-=
5.net>> wrote:

All,

I was just reviewing the proposed charter and one thing we might consider a=
dding is the open database item (formally standardized datasets). We need t=
his not only to make progress and to prove out how various approaches work =
(or don't work), whether new approaches improve on existing methods, and th=
e like.

More generally, the machine learning community has a long standing traditio=
n of open source, open data, and even open models (AlexNet, GoogLeNet, etc =
are excellent examples of the later). The NMLRG should strive to carry on t=
hese traditions, and they should probably be reflected in the charter. Perh=
aps something like:

+1 On keeping this long standing tradition.


To "The Network Machine Learning Research Group (NMLRG) provides a forum fo=
r researchers to explore the potential of machine learning technologies for=
 networks. In particular, the NMLRG will work on potential approaches that =
apply machine learning technologies in network control, network management,=
 and supplying network data for upper-layer applications.'

add something like


"In addition and to the extent possible, approaches proposed in the NMLRG s=
hould be documented not only with traditional written documentation but als=
o with an open source reference implementation and open access to the data =
used to verify the approach."

or similar, again with the goal of supporting/carrying on the open science =
approach that has been so successful to date in the ML community.

+1 again. Would it be possible to use a hosted service such as Github or th=
e like for hosting code, documentation, and data-set(s)? I think, tradition=
ally, there was some adversity to using a hosted service like Github, at le=
ast within the IETF, but in general the ML community rallies around this pl=
atform. So it would almost make sense in that respect.


Thanks,

--dmm


_______________________________________________
nmlrg mailing list
nmlrg@irtf.org<mailto:nmlrg@irtf.org>
https://www.irtf.org/mailman/listinfo/nmlrg



--_000_5D36713D8A4E7348A7E10DF7437A4B927CA89EC7NKGEML515MBXchi_
Content-Type: text/html; charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable

<html dir=3D"ltr">
<head>
<meta http-equiv=3D"Content-Type" content=3D"text/html; charset=3Diso-8859-=
1">
<style id=3D"owaParaStyle">P {
	MARGIN-TOP: 0px; MARGIN-BOTTOM: 0px
}
P {
	MARGIN-TOP: 0px; MARGIN-BOTTOM: 0px
}
</style>
</head>
<body fPStyle=3D"1" ocsi=3D"0">
<div style=3D"FONT-FAMILY: Tahoma; DIRECTION: ltr; COLOR: #000000; FONT-SIZ=
E: 10pt">
<p>Thanks David and Ed,</p>
<p>&nbsp;</p>
<p>This is indeed a good suggestion. I also like the openness although&nbsp=
;our traditional players in&nbsp;the network area, like operators and vendo=
rs, may not be as open as ML communities. So, I&nbsp;may soften requirement=
s a little&nbsp;bit.&nbsp;&nbsp;We could certainly add&nbsp;such text&nbsp;=
when
 the next time we modify the charter. There may be a chance to do so someti=
me later this year when we formally approved as a long-running IRTF researc=
h group, if we could get approval.</p>
<p>&nbsp;</p>
<p>Regards,</p>
<p>&nbsp;</p>
<p>Sheng</p>
<p>&nbsp;</p>
<div style=3D"FONT-FAMILY: Times New Roman; COLOR: #000000; FONT-SIZE: 16px=
">
<hr tabindex=3D"-1">
<div style=3D"DIRECTION: ltr" id=3D"divRpF804044"><font color=3D"#000000" s=
ize=3D"2" face=3D"Tahoma"><b>From:</b> nmlrg [nmlrg-bounces@irtf.org] on be=
half of Ed Henry [networkn3rd@gmail.com]<br>
<b>Sent:</b> 30 June 2016 4:11<br>
<b>To:</b> David Meyer<br>
<b>Cc:</b> nmlrg@irtf.org<br>
<b>Subject:</b> Re: [Nmlrg] One comment on the proposed charter...<br>
</font><br>
</div>
<div></div>
<div>
<div dir=3D"ltr"><br>
<div class=3D"gmail_extra"><br>
<div class=3D"gmail_quote">On Wed, Jun 29, 2016 at 1:59 PM, David Meyer <sp=
an dir=3D"ltr">
&lt;<a href=3D"mailto:dmm@1-4-5.net" target=3D"_blank">dmm@1-4-5.net</a>&gt=
;</span> wrote:<br>
<blockquote style=3D"BORDER-LEFT: rgb(204,204,204) 1px solid; MARGIN: 0px 0=
px 0px 0.8ex; PADDING-LEFT: 1ex" class=3D"gmail_quote">
<div dir=3D"ltr">
<div><br>
</div>
<div>All,</div>
<div><br>
</div>
<div>I was just reviewing the proposed charter and one thing we might consi=
der adding is the open database item (formally standardized datasets). We n=
eed this not only to make progress and to prove out how various approaches =
work (or don't work), whether new
 approaches improve on existing methods, and the like.</div>
<div><br>
</div>
<div>More generally, the machine learning community has a long standing tra=
dition of open source, open data, and even open models (AlexNet, GoogLeNet,=
 etc are excellent examples of the later). The NMLRG should strive to carry=
 on these traditions, and they should
 probably be reflected in the charter. Perhaps something like:</div>
</div>
</blockquote>
<div><br>
</div>
<div>&#43;1 On keeping this long standing tradition.</div>
<div><br>
</div>
<blockquote style=3D"BORDER-LEFT: rgb(204,204,204) 1px solid; MARGIN: 0px 0=
px 0px 0.8ex; PADDING-LEFT: 1ex" class=3D"gmail_quote">
<div dir=3D"ltr">
<div><br>
</div>
<div>To &quot;The Network Machine Learning Research Group (NMLRG) provides =
a forum for researchers to explore the potential of machine learning techno=
logies for networks. In particular, the NMLRG will work on potential approa=
ches that apply machine learning technologies
 in network control, network management, and supplying network data for upp=
er-layer applications.' &nbsp;</div>
<div><br>
</div>
<div>add something like</div>
<div><br>
</div>
<div><br>
</div>
<div>&quot;In addition and to the extent possible, approaches proposed in t=
he NMLRG should be documented not only with traditional written documentati=
on but also with an open source reference implementation and open access to=
 the data used to verify the approach.&quot;</div>
</div>
</blockquote>
<blockquote style=3D"BORDER-LEFT: rgb(204,204,204) 1px solid; MARGIN: 0px 0=
px 0px 0.8ex; PADDING-LEFT: 1ex" class=3D"gmail_quote">
<div dir=3D"ltr">
<div><br>
</div>
<div>or similar, again with the goal of supporting/carrying on the open sci=
ence approach that has been so successful to date in the ML community.</div=
>
</div>
</blockquote>
<div><br>
</div>
<div>&#43;1 again. Would it be possible to use a hosted service such as Git=
hub or the like for hosting code, documentation, and data-set(s)? I think, =
traditionally, there was some adversity to using a hosted service like Gith=
ub, at least within the IETF, but in
 general the ML community rallies around this platform. So it would almost =
make sense in that respect.</div>
<div>&nbsp;</div>
<blockquote style=3D"BORDER-LEFT: rgb(204,204,204) 1px solid; MARGIN: 0px 0=
px 0px 0.8ex; PADDING-LEFT: 1ex" class=3D"gmail_quote">
<div dir=3D"ltr">
<div><br>
</div>
<div>Thanks,</div>
<div><br>
</div>
<div>--dmm</div>
<div><br>
</div>
</div>
<br>
_______________________________________________<br>
nmlrg mailing list<br>
<a href=3D"mailto:nmlrg@irtf.org" target=3D"_blank">nmlrg@irtf.org</a><br>
<a href=3D"https://www.irtf.org/mailman/listinfo/nmlrg" rel=3D"noreferrer" =
target=3D"_blank">https://www.irtf.org/mailman/listinfo/nmlrg</a><br>
<br>
</blockquote>
</div>
<br>
</div>
</div>
</div>
</div>
</div>
</body>
</html>

--_000_5D36713D8A4E7348A7E10DF7437A4B927CA89EC7NKGEML515MBXchi_--


From nobody Thu Jun 30 10:45:44 2016
Return-Path: <jiangsheng@huawei.com>
X-Original-To: nmlrg@ietfa.amsl.com
Delivered-To: nmlrg@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 5498612DB3E for <nmlrg@ietfa.amsl.com>; Thu, 30 Jun 2016 10:45:42 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -5.646
X-Spam-Level: 
X-Spam-Status: No, score=-5.646 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, HTML_MESSAGE=0.001, RCVD_IN_DNSWL_MED=-2.3, RCVD_IN_MSPIKE_H3=-0.01, RCVD_IN_MSPIKE_WL=-0.01, RP_MATCHES_RCVD=-1.426, SPF_PASS=-0.001] autolearn=ham autolearn_force=no
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id rmhWkY0M9Fag for <nmlrg@ietfa.amsl.com>; Thu, 30 Jun 2016 10:45:39 -0700 (PDT)
Received: from szxga01-in.huawei.com (szxga01-in.huawei.com [58.251.152.64]) (using TLSv1 with cipher RC4-SHA (128/128 bits)) (No client certificate requested) by ietfa.amsl.com (Postfix) with ESMTPS id 47E7B12DB25 for <nmlrg@irtf.org>; Thu, 30 Jun 2016 10:45:07 -0700 (PDT)
Received: from 172.24.1.60 (EHLO nkgeml412-hub.china.huawei.com) ([172.24.1.60]) by szxrg01-dlp.huawei.com (MOS 4.3.7-GA FastPath queued) with ESMTP id DNC61289; Fri, 01 Jul 2016 01:44:45 +0800 (CST)
Received: from NKGEML515-MBX.china.huawei.com ([fe80::a54a:89d2:c471:ff]) by nkgeml412-hub.china.huawei.com ([10.98.56.73]) with mapi id 14.03.0235.001; Fri, 1 Jul 2016 01:44:38 +0800
From: Sheng Jiang <jiangsheng@huawei.com>
To: David Meyer <dmm@1-4-5.net>
Thread-Topic: [Nmlrg] links to each slide of presentations//RE: slides of NMLRG #3, June 27th, Athens
Thread-Index: AQHR0UXZfny1k5v9oUOv37+HaylYt6AAOhoAgAFEEJY=
Date: Thu, 30 Jun 2016 17:44:37 +0000
Message-ID: <5D36713D8A4E7348A7E10DF7437A4B927CA89ED9@NKGEML515-MBX.china.huawei.com>
References: <5D36713D8A4E7348A7E10DF7437A4B927CA893E9@NKGEML515-MBX.china.huawei.com>,  <CAHiKxWi_mru0CG1ysoSVvN+beadHnmat2hzo5qeAQFudeYv75w@mail.gmail.com>
In-Reply-To: <CAHiKxWi_mru0CG1ysoSVvN+beadHnmat2hzo5qeAQFudeYv75w@mail.gmail.com>
Accept-Language: en-GB, zh-CN, en-US
Content-Language: en-GB
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
x-originating-ip: [10.47.64.200]
Content-Type: multipart/alternative; boundary="_000_5D36713D8A4E7348A7E10DF7437A4B927CA89ED9NKGEML515MBXchi_"
MIME-Version: 1.0
X-CFilter-Loop: Reflected
X-Mirapoint-Virus-RAPID-Raw: score=unknown(0), refid=str=0001.0A090205.57755A8E.0037, ss=1, re=0.000, recu=0.000, reip=0.000,  cl=1, cld=1, fgs=0, ip=0.0.0.0, so=2013-06-18 04:22:30, dmn=2013-03-21 17:37:32
X-Mirapoint-Loop-Id: 5c88429eac5262f2308c82d61ca5f17c
Archived-At: <https://mailarchive.ietf.org/arch/msg/nmlrg/o48oB-54ZvRBUvDCSVJdebBAbmo>
Cc: "nmlrg@irtf.org" <nmlrg@irtf.org>, "kt@incelligent.net" <kt@incelligent.net>, =?windows-1256?Q?Panagiotis_Demestichas_=FD=5Bpdemestichas=40gmail=2Ecom?= =?windows-1256?Q?=5D=FD?= <pdemestichas@gmail.com>
Subject: Re: [Nmlrg] links to each slide of presentations//RE: slides of NMLRG #3, June 27th, Athens
X-BeenThere: nmlrg@irtf.org
X-Mailman-Version: 2.1.17
Precedence: list
List-Id: Network Machine Learning Research Group <nmlrg.irtf.org>
List-Unsubscribe: <https://www.irtf.org/mailman/options/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=unsubscribe>
List-Archive: <https://mailarchive.ietf.org/arch/browse/nmlrg/>
List-Post: <mailto:nmlrg@irtf.org>
List-Help: <mailto:nmlrg-request@irtf.org?subject=help>
List-Subscribe: <https://www.irtf.org/mailman/listinfo/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=subscribe>
X-List-Received-Date: Thu, 30 Jun 2016 17:45:42 -0000

--_000_5D36713D8A4E7348A7E10DF7437A4B927CA89ED9NKGEML515MBXchi_
Content-Type: text/plain; charset="windows-1256"
Content-Transfer-Encoding: quoted-printable

Hi, David,



I am not sure your question is for me or for the speaker. Copied this email=
 to the speaker, Panagiotis. If the question is on me, I don't really under=
stand the point of your question. Could you please provide more context or =
explanation?



Thanks,



Sheng



________________________________
From: David Meyer [dmm@1-4-5.net]
Sent: 30 June 2016 2:09
To: Sheng Jiang
Cc: nmlrg@irtf.org
Subject: Re: [Nmlrg] links to each slide of presentations//RE: slides of NM=
LRG #3, June 27th, Athens

This question is on this use case:

Mobile network state characterization and prediction
https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interi=
m-2016-nmlrg-01-4.pdf

Isn't it the case that you will know if something is registered to the MBB,=
 MMC, or MCC slice at device authentication time? If not what is the scenar=
io in which you don't know?

Thanks,

--dmm



On Tue, Jun 28, 2016 at 7:03 AM, Sheng Jiang <jiangsheng@huawei.com<mailto:=
jiangsheng@huawei.com>> wrote:
Oops... The proceeding page for interim meeting seems not be as intelligent=
 as the proceeding pages of IETF meetings. Our proceeding page does not aut=
onomically show the slides. I have sent an email to IETF secretary to ask t=
hem to fix it. Meanwhile, in this email, here is the links for each present=
ations:

Slide Filename Edit Replace Delete
Chair Slides
https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interi=
m-2016-nmlrg-01-0.pdf

Introduction to Network Machine Learning & NMLRG
https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interi=
m-2016-nmlrg-01-1.pdf

Data Collection and Analysis At High Security Lab
https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interi=
m-2016-nmlrg-01-2.pdf

Use Cases of Applying Machine Learning Mechanism with Network Traffic
https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interi=
m-2016-nmlrg-01-3.pdf

Mobile network state characterization and prediction
https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interi=
m-2016-nmlrg-01-4.pdf

Learning how to route
https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interi=
m-2016-nmlrg-01-5.pdf

Regards,

Sheng
________________________________________
From: nmlrg [nmlrg-bounces@irtf.org<mailto:nmlrg-bounces@irtf.org>] on beha=
lf of Sheng Jiang [jiangsheng@huawei.com<mailto:jiangsheng@huawei.com>]
Sent: 28 June 2016 21:03
To: nmlrg@irtf.org<mailto:nmlrg@irtf.org>
Subject: [Nmlrg] slides of NMLRG #3, June 27th, Athens

Hi, nmlrg,

All slides that have been presented in our NMLRG #3 meeting, June 27th, 201=
6, Athens, Greece, with EUCNC2016, have been uploaded. They can be accessed=
 through below link

https://www.ietf.org/proceedings/interim/2016/06/27/nmlrg/proceedings.html

Best regards,

Sheng
_______________________________________________
nmlrg mailing list
nmlrg@irtf.org<mailto:nmlrg@irtf.org>
https://www.irtf.org/mailman/listinfo/nmlrg
_______________________________________________
nmlrg mailing list
nmlrg@irtf.org<mailto:nmlrg@irtf.org>
https://www.irtf.org/mailman/listinfo/nmlrg


--_000_5D36713D8A4E7348A7E10DF7437A4B927CA89ED9NKGEML515MBXchi_
Content-Type: text/html; charset="windows-1256"
Content-Transfer-Encoding: quoted-printable

<html dir=3D"ltr">
<head>
<meta http-equiv=3D"Content-Type" content=3D"text/html; charset=3Dwindows-1=
256">
<style id=3D"owaParaStyle">P {
	MARGIN-TOP: 0px; MARGIN-BOTTOM: 0px
}
P {
	MARGIN-TOP: 0px; MARGIN-BOTTOM: 0px
}
</style>
</head>
<body fPStyle=3D"1" ocsi=3D"0">
<div style=3D"FONT-FAMILY: Tahoma; DIRECTION: ltr; COLOR: #000000; FONT-SIZ=
E: 10pt">
<p>Hi, David,</p>
<p>&nbsp;</p>
<p>I am not sure your question is for me or for the speaker. Copied this em=
ail to the speaker, Panagiotis. If the question is on me, I don't really un=
derstand the point of your question. Could you please provide more context =
or explanation?</p>
<p>&nbsp;</p>
<p>Thanks,</p>
<p>&nbsp;</p>
<p>Sheng</p>
<p>&nbsp;</p>
<div style=3D"FONT-FAMILY: Times New Roman; COLOR: #000000; FONT-SIZE: 16px=
">
<hr tabindex=3D"-1">
<div style=3D"DIRECTION: ltr" id=3D"divRpF537962"><font color=3D"#000000" s=
ize=3D"2" face=3D"Tahoma"><b>From:</b> David Meyer [dmm@1-4-5.net]<br>
<b>Sent:</b> 30 June 2016 2:09<br>
<b>To:</b> Sheng Jiang<br>
<b>Cc:</b> nmlrg@irtf.org<br>
<b>Subject:</b> Re: [Nmlrg] links to each slide of presentations//RE: slide=
s of NMLRG #3, June 27th, Athens<br>
</font><br>
</div>
<div></div>
<div>
<div dir=3D"ltr">This question is on this use case:
<div><br>
<div>
<div>Mobile network state characterization and prediction</div>
<div><a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slid=
es/slides-interim-2016-nmlrg-01-4.pdf" target=3D"_blank">https://www.ietf.o=
rg/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-4.=
pdf</a></div>
</div>
<div><br>
</div>
<div>Isn't it the case that you will know if something is registered to the=
 MBB, MMC, or MCC slice at device authentication time? If not what is the s=
cenario in which you don't know?</div>
<div><br>
</div>
<div>Thanks,</div>
<div><br>
</div>
<div>--dmm</div>
<div><br>
</div>
<div><br>
<div class=3D"gmail_extra"><br>
<div class=3D"gmail_quote">On Tue, Jun 28, 2016 at 7:03 AM, Sheng Jiang <sp=
an dir=3D"ltr">
&lt;<a href=3D"mailto:jiangsheng@huawei.com" target=3D"_blank">jiangsheng@h=
uawei.com</a>&gt;</span> wrote:<br>
<blockquote style=3D"BORDER-LEFT: rgb(204,204,204) 1px solid; MARGIN: 0px 0=
px 0px 0.8ex; PADDING-LEFT: 1ex" class=3D"gmail_quote">
Oops... The proceeding page for interim meeting seems not be as intelligent=
 as the proceeding pages of IETF meetings. Our proceeding page does not aut=
onomically show the slides. I have sent an email to IETF secretary to ask t=
hem to fix it. Meanwhile, in this
 email, here is the links for each presentations:<br>
<br>
Slide Filename Edit Replace Delete<br>
Chair Slides<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-0.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-0.pdf</a><br>
<br>
Introduction to Network Machine Learning &amp; NMLRG<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-1.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-1.pdf</a><br>
<br>
Data Collection and Analysis At High Security Lab<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-2.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-2.pdf</a><br>
<br>
Use Cases of Applying Machine Learning Mechanism with Network Traffic<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-3.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-3.pdf</a><br>
<br>
Mobile network state characterization and prediction<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-4.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-4.pdf</a><br>
<br>
Learning how to route<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-5.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-5.pdf</a><br>
<br>
Regards,<br>
<br>
Sheng<br>
________________________________________<br>
From: nmlrg [<a href=3D"mailto:nmlrg-bounces@irtf.org" target=3D"_blank">nm=
lrg-bounces@irtf.org</a>] on behalf of Sheng Jiang [<a href=3D"mailto:jiang=
sheng@huawei.com" target=3D"_blank">jiangsheng@huawei.com</a>]<br>
Sent: 28 June 2016 21:03<br>
To: <a href=3D"mailto:nmlrg@irtf.org" target=3D"_blank">nmlrg@irtf.org</a><=
br>
Subject: [Nmlrg] slides of NMLRG #3, June 27th, Athens<br>
<br>
Hi, nmlrg,<br>
<br>
All slides that have been presented in our NMLRG #3 meeting, June 27th, 201=
6, Athens, Greece, with EUCNC2016, have been uploaded. They can be accessed=
 through below link<br>
<br>
<a href=3D"https://www.ietf.org/proceedings/interim/2016/06/27/nmlrg/procee=
dings.html" rel=3D"noreferrer" target=3D"_blank">https://www.ietf.org/proce=
edings/interim/2016/06/27/nmlrg/proceedings.html</a><br>
<br>
Best regards,<br>
<br>
Sheng<br>
_______________________________________________<br>
nmlrg mailing list<br>
<a href=3D"mailto:nmlrg@irtf.org" target=3D"_blank">nmlrg@irtf.org</a><br>
<a href=3D"https://www.irtf.org/mailman/listinfo/nmlrg" rel=3D"noreferrer" =
target=3D"_blank">https://www.irtf.org/mailman/listinfo/nmlrg</a><br>
_______________________________________________<br>
nmlrg mailing list<br>
<a href=3D"mailto:nmlrg@irtf.org" target=3D"_blank">nmlrg@irtf.org</a><br>
<a href=3D"https://www.irtf.org/mailman/listinfo/nmlrg" rel=3D"noreferrer" =
target=3D"_blank">https://www.irtf.org/mailman/listinfo/nmlrg</a><br>
</blockquote>
</div>
<br>
</div>
</div>
</div>
</div>
</div>
</div>
</div>
</body>
</html>

--_000_5D36713D8A4E7348A7E10DF7437A4B927CA89ED9NKGEML515MBXchi_--


From nobody Thu Jun 30 10:57:18 2016
Return-Path: <dmm@1-4-5.net>
X-Original-To: nmlrg@ietfa.amsl.com
Delivered-To: nmlrg@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id C3BBA12DB20 for <nmlrg@ietfa.amsl.com>; Thu, 30 Jun 2016 10:57:17 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.599
X-Spam-Level: 
X-Spam-Status: No, score=-2.599 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, HTML_MESSAGE=0.001, RCVD_IN_DNSWL_LOW=-0.7] autolearn=ham autolearn_force=no
Authentication-Results: ietfa.amsl.com (amavisd-new); dkim=pass (2048-bit key) header.d=1-4-5-net.20150623.gappssmtp.com
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id pIKSihFdADzM for <nmlrg@ietfa.amsl.com>; Thu, 30 Jun 2016 10:57:15 -0700 (PDT)
Received: from mail-io0-x235.google.com (mail-io0-x235.google.com [IPv6:2607:f8b0:4001:c06::235]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by ietfa.amsl.com (Postfix) with ESMTPS id 2C15812D59F for <nmlrg@irtf.org>; Thu, 30 Jun 2016 10:57:15 -0700 (PDT)
Received: by mail-io0-x235.google.com with SMTP id g13so78316888ioj.1 for <nmlrg@irtf.org>; Thu, 30 Jun 2016 10:57:15 -0700 (PDT)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1-4-5-net.20150623.gappssmtp.com; s=20150623; h=mime-version:in-reply-to:references:from:date:message-id:subject:to :cc; bh=OreDR9Hmb5hv9L8J3h8KA4m2yBnX86MMTy1/R/Fd/as=; b=KaK08bS1DR62jEzYlnxpFd2PwtqR/jkU/jOlpCS1XvMknvi+ufg9ZQe3Trh1Wnb8kC VF6AVa5IZ2fNYZafuqW5ezPDz6E6YYC6tgN+dKT4VHpmxqa81gveI+bNsBfWcComfPu7 yOgsUKwspk+el1BTGzlihzs1JgcX7CegxonFqREZla5AVXAsvkewd/UbDqbvlyfgKb+K JHUJcNRS5E2iTXmo4R+qYI0dcO6+bt12etNXgf69zpJ1jypOU7RfACR73jwxco96KduL L7jEn8EOGdPOWMJuPSrMN2dvf2Co0K6BUBWk23Wgp4cqDKfnvYqJ2EdDMsapiQdOdTQz 6Pwg==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:from:date :message-id:subject:to:cc; bh=OreDR9Hmb5hv9L8J3h8KA4m2yBnX86MMTy1/R/Fd/as=; b=XgUE+BxN+M50o9qDT+A3VE+18yYJ6LfIedwGwZGymn4PYqCc9/xWYo+MwN5aWUMeWI mUOZk20oMMCCBDTpGfxgyZTJOvPzytmnwhZAk8WqJygkoSwisJG+K7+Nt83tUrQKXszg rAJMYphUZplbBeV+UrhTXG0gKAgzQs46cilb6Ciosd6OOGOBqpRsXFPrZfD+HWObKb2v AMjFfbHlyheHPDwletTcfFllYiaJOMnq+QFQZo4o8YLBOTpWFMP0Kw1SClu5ZK5dohRY 74xTLfmmtxHxKXs7VUDkfh3Zf8ftddM8VcnvjLdBEC8JB6DDddx+qotBBkf3DB7N0Kb3 7aKQ==
X-Gm-Message-State: ALyK8tID6FipvAcs+3EHNFH7hz+LwAamG5ey/Uqw0izU+3+0v1a4Rv54bRcukiwk2hAZQdHIW1Z+1HQ7WBjJxg==
X-Received: by 10.107.29.142 with SMTP id d136mr992345iod.50.1467309434507; Thu, 30 Jun 2016 10:57:14 -0700 (PDT)
MIME-Version: 1.0
Received: by 10.64.148.133 with HTTP; Thu, 30 Jun 2016 10:57:13 -0700 (PDT)
X-Originating-IP: [128.223.156.253]
In-Reply-To: <5D36713D8A4E7348A7E10DF7437A4B927CA89EC7@NKGEML515-MBX.china.huawei.com>
References: <CAHiKxWihP=9jQDb2X1DB3EEoNPoB+WSnmRJAU9zo-pTf9tjq=w@mail.gmail.com> <CADnDDRFrMp684SMLiGjNE_hszZ8CABG5o2tGdZp9U6_LOJYetw@mail.gmail.com> <5D36713D8A4E7348A7E10DF7437A4B927CA89EC7@NKGEML515-MBX.china.huawei.com>
From: David Meyer <dmm@1-4-5.net>
Date: Thu, 30 Jun 2016 10:57:13 -0700
Message-ID: <CAHiKxWgYo9=f8p5KBj5Y2Y2nPhJwetnyxyN5v1rvauujFHUoLA@mail.gmail.com>
To: Sheng Jiang <jiangsheng@huawei.com>
Content-Type: multipart/alternative; boundary=001a114095849eb12005368299c0
Archived-At: <https://mailarchive.ietf.org/arch/msg/nmlrg/h3dWnJOkadtWLWgm2i4pc1dXoNI>
Cc: "nmlrg@irtf.org" <nmlrg@irtf.org>, Ed Henry <networkn3rd@gmail.com>
Subject: Re: [Nmlrg] One comment on the proposed charter...
X-BeenThere: nmlrg@irtf.org
X-Mailman-Version: 2.1.17
Precedence: list
List-Id: Network Machine Learning Research Group <nmlrg.irtf.org>
List-Unsubscribe: <https://www.irtf.org/mailman/options/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=unsubscribe>
List-Archive: <https://mailarchive.ietf.org/arch/browse/nmlrg/>
List-Post: <mailto:nmlrg@irtf.org>
List-Help: <mailto:nmlrg-request@irtf.org?subject=help>
List-Subscribe: <https://www.irtf.org/mailman/listinfo/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=subscribe>
X-List-Received-Date: Thu, 30 Jun 2016 17:57:18 -0000

--001a114095849eb12005368299c0
Content-Type: text/plain; charset=UTF-8

On Thu, Jun 30, 2016 at 10:34 AM, Sheng Jiang <jiangsheng@huawei.com> wrote:

> Thanks David and Ed,
>
>
>
> This is indeed a good suggestion. I also like the openness although our
> traditional players in the network area, like operators and vendors, may
> not be as open as ML communities. So, I may soften requirements a
> little bit.  We could certainly add such text when the next time we modify
> the charter. There may be a chance to do so sometime later this year when
> we formally approved as a long-running IRTF research group, if we could get
> approval.
>
>
>

Thanks Sheng.

I'll just mention here that (a). the Internet has a long tradition of
openness, perhaps best exemplified by the IETF itself; I don't see a
compelling reason why "ML for networking" should be any different [0]. In
particular, all the same problems the exist here as existed in the early
days of networking, when close to everything was proprietary. In any event
if there is a compelling reason why we should admit closed/proprietary
solutions/proposals that should be called out explicitly (obviously I would
disagree with this as a charter item but it should be made explicit if that
is what people are thinking), and  (b). the ML environment is rapidly
changing and making progress requires collaboration and sharing of
knowledge; that usually doesn't happen around closed, proprietary solutions
(almost by definition).

Finally, since AFAIK the RG hasn't been formally chartered there will be a
charter review and revision cycle before formal chartering (at least that
has traditionally been the case).

Thx,

--dmm

[0] And again, as I mentioned in an earlier email, much of the progress we
see in the ML space derives directly or indirectly from the Open Science
tradition we see in the ML community.

Regards,
>
>
>
> Sheng
>
>
> ------------------------------
> *From:* nmlrg [nmlrg-bounces@irtf.org] on behalf of Ed Henry [
> networkn3rd@gmail.com]
> *Sent:* 30 June 2016 4:11
> *To:* David Meyer
> *Cc:* nmlrg@irtf.org
> *Subject:* Re: [Nmlrg] One comment on the proposed charter...
>
>
>
> On Wed, Jun 29, 2016 at 1:59 PM, David Meyer <dmm@1-4-5.net> wrote:
>
>>
>> All,
>>
>> I was just reviewing the proposed charter and one thing we might consider
>> adding is the open database item (formally standardized datasets). We need
>> this not only to make progress and to prove out how various approaches work
>> (or don't work), whether new approaches improve on existing methods, and
>> the like.
>>
>> More generally, the machine learning community has a long standing
>> tradition of open source, open data, and even open models (AlexNet,
>> GoogLeNet, etc are excellent examples of the later). The NMLRG should
>> strive to carry on these traditions, and they should probably be reflected
>> in the charter. Perhaps something like:
>>
>
> +1 On keeping this long standing tradition.
>
>
>> To "The Network Machine Learning Research Group (NMLRG) provides a forum
>> for researchers to explore the potential of machine learning technologies
>> for networks. In particular, the NMLRG will work on potential approaches
>> that apply machine learning technologies in network control, network
>> management, and supplying network data for upper-layer applications.'
>>
>> add something like
>>
>>
>> "In addition and to the extent possible, approaches proposed in the NMLRG
>> should be documented not only with traditional written documentation but
>> also with an open source reference implementation and open access to the
>> data used to verify the approach."
>>
>
>> or similar, again with the goal of supporting/carrying on the open
>> science approach that has been so successful to date in the ML community.
>>
>
> +1 again. Would it be possible to use a hosted service such as Github or
> the like for hosting code, documentation, and data-set(s)? I think,
> traditionally, there was some adversity to using a hosted service like
> Github, at least within the IETF, but in general the ML community rallies
> around this platform. So it would almost make sense in that respect.
>
>
>>
>> Thanks,
>>
>> --dmm
>>
>>
>> _______________________________________________
>> nmlrg mailing list
>> nmlrg@irtf.org
>> https://www.irtf.org/mailman/listinfo/nmlrg
>>
>>
>

--001a114095849eb12005368299c0
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><br><div class=3D"gmail_extra"><br><div class=3D"gmail_quo=
te">On Thu, Jun 30, 2016 at 10:34 AM, Sheng Jiang <span dir=3D"ltr">&lt;<a =
href=3D"mailto:jiangsheng@huawei.com" target=3D"_blank">jiangsheng@huawei.c=
om</a>&gt;</span> wrote:<br><blockquote class=3D"gmail_quote" style=3D"marg=
in:0px 0px 0px 0.8ex;border-left-width:1px;border-left-style:solid;border-l=
eft-color:rgb(204,204,204);padding-left:1ex">




<div>
<div style=3D"font-family:Tahoma;direction:ltr;color:rgb(0,0,0);font-size:1=
0pt">
<p>Thanks David and Ed,</p>
<p>=C2=A0</p>
<p>This is indeed a good suggestion. I also like the openness although=C2=
=A0our traditional players in=C2=A0the network area, like operators and ven=
dors, may not be as open as ML communities. So, I=C2=A0may soften requireme=
nts a little=C2=A0bit.=C2=A0=C2=A0We could certainly add=C2=A0such text=C2=
=A0when
 the next time we modify the charter. There may be a chance to do so someti=
me later this year when we formally approved as a long-running IRTF researc=
h group, if we could get approval.</p>
<p>=C2=A0</p></div></div></blockquote><div><br></div><div>Thanks Sheng.=C2=
=A0</div><div><br></div><div>I&#39;ll just mention here that (a). the Inter=
net has a long tradition of openness, perhaps best exemplified by the IETF =
itself; I don&#39;t see a compelling reason why &quot;ML for networking&quo=
t; should be any different [0]. In particular, all the same problems the ex=
ist here as existed in the early days of networking, when close to everythi=
ng was proprietary. In any event if there is a compelling reason why we sho=
uld admit closed/proprietary solutions/proposals that should be called out =
explicitly (obviously I would disagree with this as a charter item but it s=
hould be made explicit if that is what people are thinking), and =C2=A0(b).=
 the ML environment is rapidly changing and making progress requires collab=
oration and sharing of knowledge; that usually doesn&#39;t happen around cl=
osed, proprietary solutions (almost by definition).=C2=A0</div><div><br></d=
iv><div>Finally, since AFAIK the RG hasn&#39;t been formally chartered ther=
e will be a charter review and revision cycle before formal chartering (at =
least that has traditionally been the case).=C2=A0</div><div><br></div><div=
>Thx,</div><div><br></div><div>--dmm</div><div><br></div><div>[0] And again=
, as I mentioned in an earlier email, much of the progress we see in the ML=
 space derives directly or indirectly from the Open Science tradition we se=
e in the ML community.</div><div><br></div><blockquote class=3D"gmail_quote=
" style=3D"margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-style=
:solid;border-left-color:rgb(204,204,204);padding-left:1ex"><div><div style=
=3D"font-family:Tahoma;direction:ltr;color:rgb(0,0,0);font-size:10pt">
<p>Regards,</p>
<p>=C2=A0</p>
<p>Sheng</p>
<p>=C2=A0</p>
<div style=3D"font-family:&quot;Times New Roman&quot;;color:rgb(0,0,0);font=
-size:16px">
<hr>
<div style=3D"direction:ltr"><font color=3D"#000000" size=3D"2" face=3D"Tah=
oma"><b>From:</b> nmlrg [<a href=3D"mailto:nmlrg-bounces@irtf.org" target=
=3D"_blank">nmlrg-bounces@irtf.org</a>] on behalf of Ed Henry [<a href=3D"m=
ailto:networkn3rd@gmail.com" target=3D"_blank">networkn3rd@gmail.com</a>]<b=
r>
<b>Sent:</b> 30 June 2016 4:11<br>
<b>To:</b> David Meyer<br>
<b>Cc:</b> <a href=3D"mailto:nmlrg@irtf.org" target=3D"_blank">nmlrg@irtf.o=
rg</a><br>
<b>Subject:</b> Re: [Nmlrg] One comment on the proposed charter...<br>
</font><br>
</div><div><div class=3D"h5">
<div></div>
<div>
<div dir=3D"ltr"><br>
<div class=3D"gmail_extra"><br>
<div class=3D"gmail_quote">On Wed, Jun 29, 2016 at 1:59 PM, David Meyer <sp=
an dir=3D"ltr">
&lt;<a href=3D"mailto:dmm@1-4-5.net" target=3D"_blank">dmm@1-4-5.net</a>&gt=
;</span> wrote:<br>
<blockquote style=3D"border-left-width:1px;border-left-style:solid;border-l=
eft-color:rgb(204,204,204);margin:0px 0px 0px 0.8ex;padding-left:1ex" class=
=3D"gmail_quote">
<div dir=3D"ltr">
<div><br>
</div>
<div>All,</div>
<div><br>
</div>
<div>I was just reviewing the proposed charter and one thing we might consi=
der adding is the open database item (formally standardized datasets). We n=
eed this not only to make progress and to prove out how various approaches =
work (or don&#39;t work), whether new
 approaches improve on existing methods, and the like.</div>
<div><br>
</div>
<div>More generally, the machine learning community has a long standing tra=
dition of open source, open data, and even open models (AlexNet, GoogLeNet,=
 etc are excellent examples of the later). The NMLRG should strive to carry=
 on these traditions, and they should
 probably be reflected in the charter. Perhaps something like:</div>
</div>
</blockquote>
<div><br>
</div>
<div>+1 On keeping this long standing tradition.</div>
<div><br>
</div>
<blockquote style=3D"border-left-width:1px;border-left-style:solid;border-l=
eft-color:rgb(204,204,204);margin:0px 0px 0px 0.8ex;padding-left:1ex" class=
=3D"gmail_quote">
<div dir=3D"ltr">
<div><br>
</div>
<div>To &quot;The Network Machine Learning Research Group (NMLRG) provides =
a forum for researchers to explore the potential of machine learning techno=
logies for networks. In particular, the NMLRG will work on potential approa=
ches that apply machine learning technologies
 in network control, network management, and supplying network data for upp=
er-layer applications.&#39; =C2=A0</div>
<div><br>
</div>
<div>add something like</div>
<div><br>
</div>
<div><br>
</div>
<div>&quot;In addition and to the extent possible, approaches proposed in t=
he NMLRG should be documented not only with traditional written documentati=
on but also with an open source reference implementation and open access to=
 the data used to verify the approach.&quot;</div>
</div>
</blockquote>
<blockquote style=3D"border-left-width:1px;border-left-style:solid;border-l=
eft-color:rgb(204,204,204);margin:0px 0px 0px 0.8ex;padding-left:1ex" class=
=3D"gmail_quote">
<div dir=3D"ltr">
<div><br>
</div>
<div>or similar, again with the goal of supporting/carrying on the open sci=
ence approach that has been so successful to date in the ML community.</div=
>
</div>
</blockquote>
<div><br>
</div>
<div>+1 again. Would it be possible to use a hosted service such as Github =
or the like for hosting code, documentation, and data-set(s)? I think, trad=
itionally, there was some adversity to using a hosted service like Github, =
at least within the IETF, but in
 general the ML community rallies around this platform. So it would almost =
make sense in that respect.</div>
<div>=C2=A0</div>
<blockquote style=3D"border-left-width:1px;border-left-style:solid;border-l=
eft-color:rgb(204,204,204);margin:0px 0px 0px 0.8ex;padding-left:1ex" class=
=3D"gmail_quote">
<div dir=3D"ltr">
<div><br>
</div>
<div>Thanks,</div>
<div><br>
</div>
<div>--dmm</div>
<div><br>
</div>
</div>
<br>
_______________________________________________<br>
nmlrg mailing list<br>
<a href=3D"mailto:nmlrg@irtf.org" target=3D"_blank">nmlrg@irtf.org</a><br>
<a href=3D"https://www.irtf.org/mailman/listinfo/nmlrg" rel=3D"noreferrer" =
target=3D"_blank">https://www.irtf.org/mailman/listinfo/nmlrg</a><br>
<br>
</blockquote>
</div>
<br>
</div>
</div>
</div>
</div></div></div>
</div>
</div>

</blockquote></div><br></div></div>

--001a114095849eb12005368299c0--


From nobody Thu Jun 30 10:58:39 2016
Return-Path: <dmm@1-4-5.net>
X-Original-To: nmlrg@ietfa.amsl.com
Delivered-To: nmlrg@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id F12CB12D1D5 for <nmlrg@ietfa.amsl.com>; Thu, 30 Jun 2016 10:58:36 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.599
X-Spam-Level: 
X-Spam-Status: No, score=-2.599 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, HTML_MESSAGE=0.001, RCVD_IN_DNSWL_LOW=-0.7] autolearn=ham autolearn_force=no
Authentication-Results: ietfa.amsl.com (amavisd-new); dkim=pass (2048-bit key) header.d=1-4-5-net.20150623.gappssmtp.com
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 84g0Nk9lu3Q9 for <nmlrg@ietfa.amsl.com>; Thu, 30 Jun 2016 10:58:34 -0700 (PDT)
Received: from mail-io0-x22c.google.com (mail-io0-x22c.google.com [IPv6:2607:f8b0:4001:c06::22c]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by ietfa.amsl.com (Postfix) with ESMTPS id A7A0012D145 for <nmlrg@irtf.org>; Thu, 30 Jun 2016 10:58:34 -0700 (PDT)
Received: by mail-io0-x22c.google.com with SMTP id g13so78350610ioj.1 for <nmlrg@irtf.org>; Thu, 30 Jun 2016 10:58:34 -0700 (PDT)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1-4-5-net.20150623.gappssmtp.com; s=20150623; h=mime-version:in-reply-to:references:from:date:message-id:subject:to :cc; bh=dAxp8jPEFV+SpaeKgw8rPLu6Nic6aebJxuFeRj6JS7A=; b=GFtLZO0gpNDbDcz9NPmYVUjEnWGebcAtlh+sBT4peZf7Ce0goBURDOFNReo2bd8aYo EDTrEx/J5FvqFozPE47R8q5MpMn545qRH1QZoPGgSIToI7Wmk2lFnewKYF0ny9B+tM+g bSZXav354tnhA3/QiGdQGL8LY5NcOer1Yt4oPREVtkBnOT82I4Bq/Doz/GkXWRHaYFPc L9sJp+hbvJPCR+kLlnoJ//nGZ5BqxeJGlQjftHiHxjUVahY+WHwWLi5Yg3c/UV5lssnA Amj1c5HJ4/vpswoU/5G6ouh/hKLAWBXFEIW43D1nZHEMA46E+Yne8dFF/j0FMNYa0PrX iomA==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:from:date :message-id:subject:to:cc; bh=dAxp8jPEFV+SpaeKgw8rPLu6Nic6aebJxuFeRj6JS7A=; b=RYnJSTKXBBZhBh/WJ6NyLOSU0sdOHB3uCF+zhMI36Uxli5cB9NpAKq+VZ6e2NMu6UM 6GqJl6f1wVD8SfNGKtvKPvK6aI99lKVLgaz05u/uwzclLD8S+6fcQ6Zvsh2sGDhvg9CU KIGVQx+Oaz5LHeNQC5f3Xt8EclhluvvzteCJKv3Jf63CBUCLx4BD5V+G9aaNuclTHB7N nYNhqOGbzjHrtEXQ/hwzRaLk1mks+78L0P9z/Cn//8IaMRXan+1vZijhY8Gx1FRICScW zKkZpNMlGT+plfpga/Hv2M86CMgRcL8SUUPTDG3ssAE5tdE4ibcYO1CHrqAbHmLx2fI+ Pv1g==
X-Gm-Message-State: ALyK8tJrA58a8Nzk/Ncl7X4KQfKXPgNV1vdZ/DVvCXgX17CJC//zq+fGz1xM/Oq1W92YV94QX/U6TEy9MfQvUQ==
X-Received: by 10.107.165.16 with SMTP id o16mr6100764ioe.56.1467309513929; Thu, 30 Jun 2016 10:58:33 -0700 (PDT)
MIME-Version: 1.0
Received: by 10.64.148.133 with HTTP; Thu, 30 Jun 2016 10:58:32 -0700 (PDT)
X-Originating-IP: [128.223.156.253]
In-Reply-To: <5D36713D8A4E7348A7E10DF7437A4B927CA89ED9@NKGEML515-MBX.china.huawei.com>
References: <5D36713D8A4E7348A7E10DF7437A4B927CA893E9@NKGEML515-MBX.china.huawei.com> <CAHiKxWi_mru0CG1ysoSVvN+beadHnmat2hzo5qeAQFudeYv75w@mail.gmail.com> <5D36713D8A4E7348A7E10DF7437A4B927CA89ED9@NKGEML515-MBX.china.huawei.com>
From: David Meyer <dmm@1-4-5.net>
Date: Thu, 30 Jun 2016 10:58:32 -0700
Message-ID: <CAHiKxWgqqZbdAUkRbpTisjcW-K6HPOmapvMYcXtLfv5Z5X8i=A@mail.gmail.com>
To: Sheng Jiang <jiangsheng@huawei.com>
Content-Type: multipart/alternative; boundary=001a1141fb0c5a97510536829e82
Archived-At: <https://mailarchive.ietf.org/arch/msg/nmlrg/b-K8UNDK-5ZmhzanaFdRbsaspc4>
Cc: "nmlrg@irtf.org" <nmlrg@irtf.org>, "kt@incelligent.net" <kt@incelligent.net>, =?UTF-8?B?UGFuYWdpb3RpcyBEZW1lc3RpY2hhcyDigI5bcGRlbWVzdGljaGFzQGdtYWlsLmNvbV3igI4=?= <pdemestichas@gmail.com>
Subject: Re: [Nmlrg] links to each slide of presentations//RE: slides of NMLRG #3, June 27th, Athens
X-BeenThere: nmlrg@irtf.org
X-Mailman-Version: 2.1.17
Precedence: list
List-Id: Network Machine Learning Research Group <nmlrg.irtf.org>
List-Unsubscribe: <https://www.irtf.org/mailman/options/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=unsubscribe>
List-Archive: <https://mailarchive.ietf.org/arch/browse/nmlrg/>
List-Post: <mailto:nmlrg@irtf.org>
List-Help: <mailto:nmlrg-request@irtf.org?subject=help>
List-Subscribe: <https://www.irtf.org/mailman/listinfo/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=subscribe>
X-List-Received-Date: Thu, 30 Jun 2016 17:58:37 -0000

--001a1141fb0c5a97510536829e82
Content-Type: text/plain; charset=UTF-8

Thanks Sheng, I meant it for the authors, but didn't see email addresses in
the deck...

--dmm


On Thu, Jun 30, 2016 at 10:44 AM, Sheng Jiang <jiangsheng@huawei.com> wrote:

> Hi, David,
>
>
>
> I am not sure your question is for me or for the speaker. Copied this
> email to the speaker, Panagiotis. If the question is on me, I don't really
> understand the point of your question. Could you please provide more
> context or explanation?
>
>
>
> Thanks,
>
>
>
> Sheng
>
>
> ------------------------------
> *From:* David Meyer [dmm@1-4-5.net]
> *Sent:* 30 June 2016 2:09
> *To:* Sheng Jiang
> *Cc:* nmlrg@irtf.org
> *Subject:* Re: [Nmlrg] links to each slide of presentations//RE: slides
> of NMLRG #3, June 27th, Athens
>
> This question is on this use case:
>
> Mobile network state characterization and prediction
>
> https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-4.pdf
>
> Isn't it the case that you will know if something is registered to the
> MBB, MMC, or MCC slice at device authentication time? If not what is the
> scenario in which you don't know?
>
> Thanks,
>
> --dmm
>
>
>
> On Tue, Jun 28, 2016 at 7:03 AM, Sheng Jiang <jiangsheng@huawei.com>
> wrote:
>
>> Oops... The proceeding page for interim meeting seems not be as
>> intelligent as the proceeding pages of IETF meetings. Our proceeding page
>> does not autonomically show the slides. I have sent an email to IETF
>> secretary to ask them to fix it. Meanwhile, in this email, here is the
>> links for each presentations:
>>
>> Slide Filename Edit Replace Delete
>> Chair Slides
>>
>> https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-0.pdf
>>
>> Introduction to Network Machine Learning & NMLRG
>>
>> https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-1.pdf
>>
>> Data Collection and Analysis At High Security Lab
>>
>> https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-2.pdf
>>
>> Use Cases of Applying Machine Learning Mechanism with Network Traffic
>>
>> https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-3.pdf
>>
>> Mobile network state characterization and prediction
>>
>> https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-4.pdf
>>
>> Learning how to route
>>
>> https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-5.pdf
>>
>> Regards,
>>
>> Sheng
>> ________________________________________
>> From: nmlrg [nmlrg-bounces@irtf.org] on behalf of Sheng Jiang [
>> jiangsheng@huawei.com]
>> Sent: 28 June 2016 21:03
>> To: nmlrg@irtf.org
>> Subject: [Nmlrg] slides of NMLRG #3, June 27th, Athens
>>
>> Hi, nmlrg,
>>
>> All slides that have been presented in our NMLRG #3 meeting, June 27th,
>> 2016, Athens, Greece, with EUCNC2016, have been uploaded. They can be
>> accessed through below link
>>
>> https://www.ietf.org/proceedings/interim/2016/06/27/nmlrg/proceedings.html
>>
>> Best regards,
>>
>> Sheng
>> _______________________________________________
>> nmlrg mailing list
>> nmlrg@irtf.org
>> https://www.irtf.org/mailman/listinfo/nmlrg
>> _______________________________________________
>> nmlrg mailing list
>> nmlrg@irtf.org
>> https://www.irtf.org/mailman/listinfo/nmlrg
>>
>
>

--001a1141fb0c5a97510536829e82
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr">Thanks Sheng, I meant it for the authors, but didn&#39;t s=
ee email addresses in the deck...<div><br></div><div>--dmm</div><div><br><d=
iv class=3D"gmail_extra"><br><div class=3D"gmail_quote">On Thu, Jun 30, 201=
6 at 10:44 AM, Sheng Jiang <span dir=3D"ltr">&lt;<a href=3D"mailto:jiangshe=
ng@huawei.com" target=3D"_blank">jiangsheng@huawei.com</a>&gt;</span> wrote=
:<br><blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-le=
ft:1px #ccc solid;padding-left:1ex">




<div>
<div style=3D"FONT-FAMILY:Tahoma;DIRECTION:ltr;COLOR:#000000;FONT-SIZE:10pt=
">
<p>Hi, David,</p>
<p>=C2=A0</p>
<p>I am not sure your question is for me or for the speaker. Copied this em=
ail to the speaker, Panagiotis. If the question is on me, I don&#39;t reall=
y understand the point of your question. Could you please provide more cont=
ext or explanation?</p>
<p>=C2=A0</p>
<p>Thanks,</p>
<p>=C2=A0</p>
<p>Sheng</p>
<p>=C2=A0</p>
<div style=3D"FONT-FAMILY:Times New Roman;COLOR:#000000;FONT-SIZE:16px">
<hr>
<div style=3D"DIRECTION:ltr"><font color=3D"#000000" size=3D"2" face=3D"Tah=
oma"><b>From:</b> David Meyer [<a href=3D"mailto:dmm@1-4-5.net" target=3D"_=
blank">dmm@1-4-5.net</a>]<br>
<b>Sent:</b> 30 June 2016 2:09<span class=3D""><br>
<b>To:</b> Sheng Jiang<br>
<b>Cc:</b> <a href=3D"mailto:nmlrg@irtf.org" target=3D"_blank">nmlrg@irtf.o=
rg</a><br>
</span><b>Subject:</b> Re: [Nmlrg] links to each slide of presentations//RE=
: slides of NMLRG #3, June 27th, Athens<br>
</font><br>
</div><div><div class=3D"h5">
<div></div>
<div>
<div dir=3D"ltr">This question is on this use case:
<div><br>
<div>
<div>Mobile network state characterization and prediction</div>
<div><a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slid=
es/slides-interim-2016-nmlrg-01-4.pdf" target=3D"_blank">https://www.ietf.o=
rg/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-4.=
pdf</a></div>
</div>
<div><br>
</div>
<div>Isn&#39;t it the case that you will know if something is registered to=
 the MBB, MMC, or MCC slice at device authentication time? If not what is t=
he scenario in which you don&#39;t know?</div>
<div><br>
</div>
<div>Thanks,</div>
<div><br>
</div>
<div>--dmm</div>
<div><br>
</div>
<div><br>
<div class=3D"gmail_extra"><br>
<div class=3D"gmail_quote">On Tue, Jun 28, 2016 at 7:03 AM, Sheng Jiang <sp=
an dir=3D"ltr">
&lt;<a href=3D"mailto:jiangsheng@huawei.com" target=3D"_blank">jiangsheng@h=
uawei.com</a>&gt;</span> wrote:<br>
<blockquote style=3D"BORDER-LEFT:rgb(204,204,204) 1px solid;MARGIN:0px 0px =
0px 0.8ex;PADDING-LEFT:1ex" class=3D"gmail_quote">
Oops... The proceeding page for interim meeting seems not be as intelligent=
 as the proceeding pages of IETF meetings. Our proceeding page does not aut=
onomically show the slides. I have sent an email to IETF secretary to ask t=
hem to fix it. Meanwhile, in this
 email, here is the links for each presentations:<br>
<br>
Slide Filename Edit Replace Delete<br>
Chair Slides<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-0.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-0.pdf</a><br>
<br>
Introduction to Network Machine Learning &amp; NMLRG<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-1.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-1.pdf</a><br>
<br>
Data Collection and Analysis At High Security Lab<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-2.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-2.pdf</a><br>
<br>
Use Cases of Applying Machine Learning Mechanism with Network Traffic<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-3.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-3.pdf</a><br>
<br>
Mobile network state characterization and prediction<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-4.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-4.pdf</a><br>
<br>
Learning how to route<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-5.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-5.pdf</a><br>
<br>
Regards,<br>
<br>
Sheng<br>
________________________________________<br>
From: nmlrg [<a href=3D"mailto:nmlrg-bounces@irtf.org" target=3D"_blank">nm=
lrg-bounces@irtf.org</a>] on behalf of Sheng Jiang [<a href=3D"mailto:jiang=
sheng@huawei.com" target=3D"_blank">jiangsheng@huawei.com</a>]<br>
Sent: 28 June 2016 21:03<br>
To: <a href=3D"mailto:nmlrg@irtf.org" target=3D"_blank">nmlrg@irtf.org</a><=
br>
Subject: [Nmlrg] slides of NMLRG #3, June 27th, Athens<br>
<br>
Hi, nmlrg,<br>
<br>
All slides that have been presented in our NMLRG #3 meeting, June 27th, 201=
6, Athens, Greece, with EUCNC2016, have been uploaded. They can be accessed=
 through below link<br>
<br>
<a href=3D"https://www.ietf.org/proceedings/interim/2016/06/27/nmlrg/procee=
dings.html" rel=3D"noreferrer" target=3D"_blank">https://www.ietf.org/proce=
edings/interim/2016/06/27/nmlrg/proceedings.html</a><br>
<br>
Best regards,<br>
<br>
Sheng<br>
_______________________________________________<br>
nmlrg mailing list<br>
<a href=3D"mailto:nmlrg@irtf.org" target=3D"_blank">nmlrg@irtf.org</a><br>
<a href=3D"https://www.irtf.org/mailman/listinfo/nmlrg" rel=3D"noreferrer" =
target=3D"_blank">https://www.irtf.org/mailman/listinfo/nmlrg</a><br>
_______________________________________________<br>
nmlrg mailing list<br>
<a href=3D"mailto:nmlrg@irtf.org" target=3D"_blank">nmlrg@irtf.org</a><br>
<a href=3D"https://www.irtf.org/mailman/listinfo/nmlrg" rel=3D"noreferrer" =
target=3D"_blank">https://www.irtf.org/mailman/listinfo/nmlrg</a><br>
</blockquote>
</div>
<br>
</div>
</div>
</div>
</div>
</div>
</div></div></div>
</div>
</div>

</blockquote></div><br></div></div></div>

--001a1141fb0c5a97510536829e82--


From nobody Thu Jun 30 11:29:16 2016
Return-Path: <pdemestichas@gmail.com>
X-Original-To: nmlrg@ietfa.amsl.com
Delivered-To: nmlrg@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id EBAFD12DB2B for <nmlrg@ietfa.amsl.com>; Thu, 30 Jun 2016 11:29:12 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.699
X-Spam-Level: 
X-Spam-Status: No, score=-2.699 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FREEMAIL_FROM=0.001, HTML_MESSAGE=0.001, RCVD_IN_DNSWL_LOW=-0.7, SPF_PASS=-0.001] autolearn=ham autolearn_force=no
Authentication-Results: ietfa.amsl.com (amavisd-new); dkim=pass (2048-bit key) header.d=gmail.com
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id GnB2EaYhT6F8 for <nmlrg@ietfa.amsl.com>; Thu, 30 Jun 2016 11:29:09 -0700 (PDT)
Received: from mail-oi0-x229.google.com (mail-oi0-x229.google.com [IPv6:2607:f8b0:4003:c06::229]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by ietfa.amsl.com (Postfix) with ESMTPS id EFD7212DAC2 for <nmlrg@irtf.org>; Thu, 30 Jun 2016 11:29:08 -0700 (PDT)
Received: by mail-oi0-x229.google.com with SMTP id u201so77626144oie.0 for <nmlrg@irtf.org>; Thu, 30 Jun 2016 11:29:08 -0700 (PDT)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20120113;  h=mime-version:in-reply-to:references:from:date:message-id:subject:to :cc; bh=mX2CYRUEhrg9nUtQ2h9wKTaIUgQYjQuKzzViGJg7tC8=; b=IkdJsOC9zC5v+kpM04liKX7LxE9Mi1b+rPJDmsvAQhRYZRAHlXCC1zTpyt3CkBwfUR 4tCJ1/dzaY8f7lXraY3ZUajsCyX5L9SeetCVvNn2/VqVleuvzs++HBB3au5HHwtMU3vC tNalNg2xYYtl61VAz5J1VX6kTYAKkM4SnHai7YT8Nc3ZGvA6U/Xgk6CcIopC82sEYMQ2 9GS8Aaua5aDkrTt3/BD7+QDa7/2sSkS/DAoPVbc4FPh2UMNJF25myR6sAiIR+vPcWgSl r1oyMy7gvyr/u2dlzxctB64a7PiFEq5ZK0PMEV8sv3k5EcdQF8GfM8CE0Rvm5OPZU6I6 IZfg==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:from:date :message-id:subject:to:cc; bh=mX2CYRUEhrg9nUtQ2h9wKTaIUgQYjQuKzzViGJg7tC8=; b=WRYcIX6j8zx51kQDIt/fA784ZzRWdcrxdqP2oecy+SI+JvL96KJGUyTJar8cBYjoTl BeI016cB4YndHpalpHK5vojQUv/B/KNBgwqyajkI9mGW+b1MI6/Kb01FsxJtTqthf75B a9HoyA1baj388q5JOEqqLSmalBdHMdWwGCWK0JfNi0mXJ9/KAbhcehb+AldRQ1H3XUcg MiwuhHXzd44p2D1WTjysCees6ozxEl1+6zAGi2UXYDgr25Tg5r+UdWJKgiNiGuItGfqH GWw79knV6ZgtJFWMCCGfZmrC6X+Iia1xzyCJM55zgYztAFvK2v/GQrr0TX4rhsEcL7Gi SpcA==
X-Gm-Message-State: ALyK8tKufFKXSNW1+6NXtJ6KCiFl0nUAzJ7YKS9dsVZfqYBqKVnbnJOa/2jwRdc6CfbBd4vDbe2AegGY87rWjA==
X-Received: by 10.202.117.144 with SMTP id q138mr6409936oic.63.1467311348276;  Thu, 30 Jun 2016 11:29:08 -0700 (PDT)
MIME-Version: 1.0
Received: by 10.157.2.68 with HTTP; Thu, 30 Jun 2016 11:28:38 -0700 (PDT)
In-Reply-To: <CAHiKxWgqqZbdAUkRbpTisjcW-K6HPOmapvMYcXtLfv5Z5X8i=A@mail.gmail.com>
References: <5D36713D8A4E7348A7E10DF7437A4B927CA893E9@NKGEML515-MBX.china.huawei.com> <CAHiKxWi_mru0CG1ysoSVvN+beadHnmat2hzo5qeAQFudeYv75w@mail.gmail.com> <5D36713D8A4E7348A7E10DF7437A4B927CA89ED9@NKGEML515-MBX.china.huawei.com> <CAHiKxWgqqZbdAUkRbpTisjcW-K6HPOmapvMYcXtLfv5Z5X8i=A@mail.gmail.com>
From: Panagiotis Demestichas <pdemestichas@gmail.com>
Date: Thu, 30 Jun 2016 21:28:38 +0300
Message-ID: <CAKo0eZOgVEoejBAyMz44iANPD=2eGKvfOS7QCp6U7S_YUTUdew@mail.gmail.com>
To: David Meyer <dmm@1-4-5.net>
Content-Type: multipart/alternative; boundary=001a11c17ad4b062f50536830ba0
Archived-At: <https://mailarchive.ietf.org/arch/msg/nmlrg/Ocm53FNq_jh8UeZUHmO9-t5iJF0>
Cc: "nmlrg@irtf.org" <nmlrg@irtf.org>, "kt@incelligent.net" <kt@incelligent.net>, Sheng Jiang <jiangsheng@huawei.com>
Subject: Re: [Nmlrg] links to each slide of presentations//RE: slides of NMLRG #3, June 27th, Athens
X-BeenThere: nmlrg@irtf.org
X-Mailman-Version: 2.1.17
Precedence: list
List-Id: Network Machine Learning Research Group <nmlrg.irtf.org>
List-Unsubscribe: <https://www.irtf.org/mailman/options/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=unsubscribe>
List-Archive: <https://mailarchive.ietf.org/arch/browse/nmlrg/>
List-Post: <mailto:nmlrg@irtf.org>
List-Help: <mailto:nmlrg-request@irtf.org?subject=help>
List-Subscribe: <https://www.irtf.org/mailman/listinfo/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=subscribe>
X-List-Received-Date: Thu, 30 Jun 2016 18:29:13 -0000

--001a11c17ad4b062f50536830ba0
Content-Type: text/plain; charset=UTF-8

Dear all

just saw the stream of emails.

The goal of the work is to learn traffic patterns, and other associated
aspects, e.g., in time and space. The goal is not to instantaneously
identify the class of the stream; such streams are used for fine tuning the
learning.

Voula Vassaki (did the presentation) will complement.

Best regards
Panagiotis



On Thu, Jun 30, 2016 at 8:58 PM, David Meyer <dmm@1-4-5.net> wrote:

> Thanks Sheng, I meant it for the authors, but didn't see email addresses
> in the deck...
>
> --dmm
>
>
> On Thu, Jun 30, 2016 at 10:44 AM, Sheng Jiang <jiangsheng@huawei.com>
> wrote:
>
>> Hi, David,
>>
>>
>>
>> I am not sure your question is for me or for the speaker. Copied this
>> email to the speaker, Panagiotis. If the question is on me, I don't really
>> understand the point of your question. Could you please provide more
>> context or explanation?
>>
>>
>>
>> Thanks,
>>
>>
>>
>> Sheng
>>
>>
>> ------------------------------
>> *From:* David Meyer [dmm@1-4-5.net]
>> *Sent:* 30 June 2016 2:09
>> *To:* Sheng Jiang
>> *Cc:* nmlrg@irtf.org
>> *Subject:* Re: [Nmlrg] links to each slide of presentations//RE: slides
>> of NMLRG #3, June 27th, Athens
>>
>> This question is on this use case:
>>
>> Mobile network state characterization and prediction
>>
>> https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-4.pdf
>>
>> Isn't it the case that you will know if something is registered to the
>> MBB, MMC, or MCC slice at device authentication time? If not what is the
>> scenario in which you don't know?
>>
>> Thanks,
>>
>> --dmm
>>
>>
>>
>> On Tue, Jun 28, 2016 at 7:03 AM, Sheng Jiang <jiangsheng@huawei.com>
>> wrote:
>>
>>> Oops... The proceeding page for interim meeting seems not be as
>>> intelligent as the proceeding pages of IETF meetings. Our proceeding page
>>> does not autonomically show the slides. I have sent an email to IETF
>>> secretary to ask them to fix it. Meanwhile, in this email, here is the
>>> links for each presentations:
>>>
>>> Slide Filename Edit Replace Delete
>>> Chair Slides
>>>
>>> https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-0.pdf
>>>
>>> Introduction to Network Machine Learning & NMLRG
>>>
>>> https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-1.pdf
>>>
>>> Data Collection and Analysis At High Security Lab
>>>
>>> https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-2.pdf
>>>
>>> Use Cases of Applying Machine Learning Mechanism with Network Traffic
>>>
>>> https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-3.pdf
>>>
>>> Mobile network state characterization and prediction
>>>
>>> https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-4.pdf
>>>
>>> Learning how to route
>>>
>>> https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-5.pdf
>>>
>>> Regards,
>>>
>>> Sheng
>>> ________________________________________
>>> From: nmlrg [nmlrg-bounces@irtf.org] on behalf of Sheng Jiang [
>>> jiangsheng@huawei.com]
>>> Sent: 28 June 2016 21:03
>>> To: nmlrg@irtf.org
>>> Subject: [Nmlrg] slides of NMLRG #3, June 27th, Athens
>>>
>>> Hi, nmlrg,
>>>
>>> All slides that have been presented in our NMLRG #3 meeting, June 27th,
>>> 2016, Athens, Greece, with EUCNC2016, have been uploaded. They can be
>>> accessed through below link
>>>
>>>
>>> https://www.ietf.org/proceedings/interim/2016/06/27/nmlrg/proceedings.html
>>>
>>> Best regards,
>>>
>>> Sheng
>>> _______________________________________________
>>> nmlrg mailing list
>>> nmlrg@irtf.org
>>> https://www.irtf.org/mailman/listinfo/nmlrg
>>> _______________________________________________
>>> nmlrg mailing list
>>> nmlrg@irtf.org
>>> https://www.irtf.org/mailman/listinfo/nmlrg
>>>
>>
>>
>

--001a11c17ad4b062f50536830ba0
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr">Dear all<br><div><br></div><div>just saw the stream of ema=
ils.</div><div><br></div><div>The goal of the work is to learn traffic patt=
erns, and other associated aspects, e.g., in time and space. The goal is no=
t to instantaneously identify the class of the stream; such streams are use=
d for fine tuning the learning.</div><div><br></div><div><div>Voula Vassaki=
 (did the presentation) will complement.</div></div><div><br></div><div>Bes=
t regards</div><div>Panagiotis</div><div><br></div><div><br></div></div><di=
v class=3D"gmail_extra"><br><div class=3D"gmail_quote">On Thu, Jun 30, 2016=
 at 8:58 PM, David Meyer <span dir=3D"ltr">&lt;<a href=3D"mailto:dmm@1-4-5.=
net" target=3D"_blank">dmm@1-4-5.net</a>&gt;</span> wrote:<br><blockquote c=
lass=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1px #ccc solid;=
padding-left:1ex"><div dir=3D"ltr">Thanks Sheng, I meant it for the authors=
, but didn&#39;t see email addresses in the deck...<div><br></div><div>--dm=
m</div><div><br><div class=3D"gmail_extra"><br><div class=3D"gmail_quote">O=
n Thu, Jun 30, 2016 at 10:44 AM, Sheng Jiang <span dir=3D"ltr">&lt;<a href=
=3D"mailto:jiangsheng@huawei.com" target=3D"_blank">jiangsheng@huawei.com</=
a>&gt;</span> wrote:<br><blockquote class=3D"gmail_quote" style=3D"margin:0=
 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">




<div>
<div style=3D"FONT-FAMILY:Tahoma;DIRECTION:ltr;COLOR:#000000;FONT-SIZE:10pt=
">
<p>Hi, David,</p>
<p>=C2=A0</p>
<p>I am not sure your question is for me or for the speaker. Copied this em=
ail to the speaker, Panagiotis. If the question is on me, I don&#39;t reall=
y understand the point of your question. Could you please provide more cont=
ext or explanation?</p>
<p>=C2=A0</p>
<p>Thanks,</p>
<p>=C2=A0</p>
<p>Sheng</p>
<p>=C2=A0</p>
<div style=3D"FONT-FAMILY:Times New Roman;COLOR:#000000;FONT-SIZE:16px">
<hr>
<div style=3D"DIRECTION:ltr"><font color=3D"#000000" size=3D"2" face=3D"Tah=
oma"><b>From:</b> David Meyer [<a href=3D"mailto:dmm@1-4-5.net" target=3D"_=
blank">dmm@1-4-5.net</a>]<br>
<b>Sent:</b> 30 June 2016 2:09<span><br>
<b>To:</b> Sheng Jiang<br>
<b>Cc:</b> <a href=3D"mailto:nmlrg@irtf.org" target=3D"_blank">nmlrg@irtf.o=
rg</a><br>
</span><b>Subject:</b> Re: [Nmlrg] links to each slide of presentations//RE=
: slides of NMLRG #3, June 27th, Athens<br>
</font><br>
</div><div><div>
<div></div>
<div>
<div dir=3D"ltr">This question is on this use case:
<div><br>
<div>
<div>Mobile network state characterization and prediction</div>
<div><a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slid=
es/slides-interim-2016-nmlrg-01-4.pdf" target=3D"_blank">https://www.ietf.o=
rg/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-4.=
pdf</a></div>
</div>
<div><br>
</div>
<div>Isn&#39;t it the case that you will know if something is registered to=
 the MBB, MMC, or MCC slice at device authentication time? If not what is t=
he scenario in which you don&#39;t know?</div>
<div><br>
</div>
<div>Thanks,</div>
<div><br>
</div>
<div>--dmm</div>
<div><br>
</div>
<div><br>
<div class=3D"gmail_extra"><br>
<div class=3D"gmail_quote">On Tue, Jun 28, 2016 at 7:03 AM, Sheng Jiang <sp=
an dir=3D"ltr">
&lt;<a href=3D"mailto:jiangsheng@huawei.com" target=3D"_blank">jiangsheng@h=
uawei.com</a>&gt;</span> wrote:<br>
<blockquote style=3D"BORDER-LEFT:rgb(204,204,204) 1px solid;MARGIN:0px 0px =
0px 0.8ex;PADDING-LEFT:1ex" class=3D"gmail_quote">
Oops... The proceeding page for interim meeting seems not be as intelligent=
 as the proceeding pages of IETF meetings. Our proceeding page does not aut=
onomically show the slides. I have sent an email to IETF secretary to ask t=
hem to fix it. Meanwhile, in this
 email, here is the links for each presentations:<br>
<br>
Slide Filename Edit Replace Delete<br>
Chair Slides<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-0.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-0.pdf</a><br>
<br>
Introduction to Network Machine Learning &amp; NMLRG<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-1.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-1.pdf</a><br>
<br>
Data Collection and Analysis At High Security Lab<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-2.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-2.pdf</a><br>
<br>
Use Cases of Applying Machine Learning Mechanism with Network Traffic<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-3.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-3.pdf</a><br>
<br>
Mobile network state characterization and prediction<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-4.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-4.pdf</a><br>
<br>
Learning how to route<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-5.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-5.pdf</a><br>
<br>
Regards,<br>
<br>
Sheng<br>
________________________________________<br>
From: nmlrg [<a href=3D"mailto:nmlrg-bounces@irtf.org" target=3D"_blank">nm=
lrg-bounces@irtf.org</a>] on behalf of Sheng Jiang [<a href=3D"mailto:jiang=
sheng@huawei.com" target=3D"_blank">jiangsheng@huawei.com</a>]<br>
Sent: 28 June 2016 21:03<br>
To: <a href=3D"mailto:nmlrg@irtf.org" target=3D"_blank">nmlrg@irtf.org</a><=
br>
Subject: [Nmlrg] slides of NMLRG #3, June 27th, Athens<br>
<br>
Hi, nmlrg,<br>
<br>
All slides that have been presented in our NMLRG #3 meeting, June 27th, 201=
6, Athens, Greece, with EUCNC2016, have been uploaded. They can be accessed=
 through below link<br>
<br>
<a href=3D"https://www.ietf.org/proceedings/interim/2016/06/27/nmlrg/procee=
dings.html" rel=3D"noreferrer" target=3D"_blank">https://www.ietf.org/proce=
edings/interim/2016/06/27/nmlrg/proceedings.html</a><br>
<br>
Best regards,<br>
<br>
Sheng<br>
_______________________________________________<br>
nmlrg mailing list<br>
<a href=3D"mailto:nmlrg@irtf.org" target=3D"_blank">nmlrg@irtf.org</a><br>
<a href=3D"https://www.irtf.org/mailman/listinfo/nmlrg" rel=3D"noreferrer" =
target=3D"_blank">https://www.irtf.org/mailman/listinfo/nmlrg</a><br>
_______________________________________________<br>
nmlrg mailing list<br>
<a href=3D"mailto:nmlrg@irtf.org" target=3D"_blank">nmlrg@irtf.org</a><br>
<a href=3D"https://www.irtf.org/mailman/listinfo/nmlrg" rel=3D"noreferrer" =
target=3D"_blank">https://www.irtf.org/mailman/listinfo/nmlrg</a><br>
</blockquote>
</div>
<br>
</div>
</div>
</div>
</div>
</div>
</div></div></div>
</div>
</div>

</blockquote></div><br></div></div></div>
</blockquote></div><br></div>

--001a11c17ad4b062f50536830ba0--


From nobody Thu Jun 30 11:39:48 2016
Return-Path: <dmm@1-4-5.net>
X-Original-To: nmlrg@ietfa.amsl.com
Delivered-To: nmlrg@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 2122512D500 for <nmlrg@ietfa.amsl.com>; Thu, 30 Jun 2016 11:39:47 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.599
X-Spam-Level: 
X-Spam-Status: No, score=-2.599 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, HTML_MESSAGE=0.001, RCVD_IN_DNSWL_LOW=-0.7] autolearn=ham autolearn_force=no
Authentication-Results: ietfa.amsl.com (amavisd-new); dkim=pass (2048-bit key) header.d=1-4-5-net.20150623.gappssmtp.com
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id S5mXGX_CslPP for <nmlrg@ietfa.amsl.com>; Thu, 30 Jun 2016 11:39:45 -0700 (PDT)
Received: from mail-it0-x22b.google.com (mail-it0-x22b.google.com [IPv6:2607:f8b0:4001:c0b::22b]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by ietfa.amsl.com (Postfix) with ESMTPS id E697812B048 for <nmlrg@irtf.org>; Thu, 30 Jun 2016 11:39:44 -0700 (PDT)
Received: by mail-it0-x22b.google.com with SMTP id g127so155191560ith.0 for <nmlrg@irtf.org>; Thu, 30 Jun 2016 11:39:44 -0700 (PDT)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1-4-5-net.20150623.gappssmtp.com; s=20150623; h=mime-version:in-reply-to:references:from:date:message-id:subject:to :cc; bh=vO5wycZXB2qipExoBxldVVqpy/KnhD0hTdyVXDzGnSI=; b=oa+R3TfzINY8e3I5bHfwqork6zRs6ck+mj+pbwfjFdOhX0+H163HNtECI2iHgoS+Jm 4ybYJIw9/jKNs5EOeehPh+C50nQ29ZC/jhxQEh0ap0HCcnuNGPpytviXiTYFwfJllQB4 /l3r+QAH71LhVipsddbAv4UFNV8pIVCqz4SO6z5bTGkYtUlqqmj4eOU/9XbHCt1QPPCu HQ4Tp581vcgp0m/kVg7Qo061ghJASQX9hIOq+XJnJ31fZHJlg0ZS9Lm2vgatQRvpE4KX Xv1q+642yicTujEZsfgEbNKOvYqWYvESKLGJBOe+0gSf+vq/jVTBSetQvEG+/WW3+zBn He0A==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:from:date :message-id:subject:to:cc; bh=vO5wycZXB2qipExoBxldVVqpy/KnhD0hTdyVXDzGnSI=; b=FvvtxRjDQPXfi5TtxO+RViTSjCWnHw9V17mjGsDDESjGZ5JX53rL6/fDfSzQ0jSqFJ TDgyNx/E3YtNK34a5x5pApU9CWHEBjT8Uo1a/tSbc2iJgD+fqcsrIAgZfUMtsen4qYRa 3KRkfQBLP15LW5kX7DXMPfISiTng9GkW0oQxvCHzWo+xGkHJsPZr7uUbJqaedI54DhTc d3I2KNyX3VmMa+HLFCfEeBUbQhsm2nj2z9elkimBin1wjqbVQgDTjT1Wol/lX9m1SmOY ZaCmCQqjO8TOcKbC0ilZdwNAumPPzx8TesrgL/EhiwAhjGC0aVaPhPzDNJWArDOcHF6h EidA==
X-Gm-Message-State: ALyK8tIWfkwgRixpkaU6h7MzfexR9fz9cD7tDK0+aRSe7KgLtmtFfdpTOvS6/U/lfiylJQvQYIM7lr+u13VfnA==
X-Received: by 10.36.227.67 with SMTP id d64mr16921650ith.18.1467311984195; Thu, 30 Jun 2016 11:39:44 -0700 (PDT)
MIME-Version: 1.0
Received: by 10.64.148.133 with HTTP; Thu, 30 Jun 2016 11:39:42 -0700 (PDT)
X-Originating-IP: [128.223.156.253]
In-Reply-To: <CAKo0eZOgVEoejBAyMz44iANPD=2eGKvfOS7QCp6U7S_YUTUdew@mail.gmail.com>
References: <5D36713D8A4E7348A7E10DF7437A4B927CA893E9@NKGEML515-MBX.china.huawei.com> <CAHiKxWi_mru0CG1ysoSVvN+beadHnmat2hzo5qeAQFudeYv75w@mail.gmail.com> <5D36713D8A4E7348A7E10DF7437A4B927CA89ED9@NKGEML515-MBX.china.huawei.com> <CAHiKxWgqqZbdAUkRbpTisjcW-K6HPOmapvMYcXtLfv5Z5X8i=A@mail.gmail.com> <CAKo0eZOgVEoejBAyMz44iANPD=2eGKvfOS7QCp6U7S_YUTUdew@mail.gmail.com>
From: David Meyer <dmm@1-4-5.net>
Date: Thu, 30 Jun 2016 11:39:42 -0700
Message-ID: <CAHiKxWihckQYL9NhZmPXC+pZymT01ZLnUY2ee4H=oeP15+K=EQ@mail.gmail.com>
To: Panagiotis Demestichas <pdemestichas@gmail.com>
Content-Type: multipart/alternative; boundary=94eb2c111ada98011205368331fd
Archived-At: <https://mailarchive.ietf.org/arch/msg/nmlrg/McapTMEwVee8jarZQz0ZJtC7r6k>
Cc: "nmlrg@irtf.org" <nmlrg@irtf.org>, "kt@incelligent.net" <kt@incelligent.net>, Sheng Jiang <jiangsheng@huawei.com>
Subject: Re: [Nmlrg] links to each slide of presentations//RE: slides of NMLRG #3, June 27th, Athens
X-BeenThere: nmlrg@irtf.org
X-Mailman-Version: 2.1.17
Precedence: list
List-Id: Network Machine Learning Research Group <nmlrg.irtf.org>
List-Unsubscribe: <https://www.irtf.org/mailman/options/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=unsubscribe>
List-Archive: <https://mailarchive.ietf.org/arch/browse/nmlrg/>
List-Post: <mailto:nmlrg@irtf.org>
List-Help: <mailto:nmlrg-request@irtf.org?subject=help>
List-Subscribe: <https://www.irtf.org/mailman/listinfo/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=subscribe>
X-List-Received-Date: Thu, 30 Jun 2016 18:39:47 -0000

--94eb2c111ada98011205368331fd
Content-Type: text/plain; charset=UTF-8

Thanks. Do you folks have a paper or other additional description of your
data and methods?

Thx,

--dmm


On Thu, Jun 30, 2016 at 11:28 AM, Panagiotis Demestichas <
pdemestichas@gmail.com> wrote:

> Dear all
>
> just saw the stream of emails.
>
> The goal of the work is to learn traffic patterns, and other associated
> aspects, e.g., in time and space. The goal is not to instantaneously
> identify the class of the stream; such streams are used for fine tuning the
> learning.
>
> Voula Vassaki (did the presentation) will complement.
>
> Best regards
> Panagiotis
>
>
>
> On Thu, Jun 30, 2016 at 8:58 PM, David Meyer <dmm@1-4-5.net> wrote:
>
>> Thanks Sheng, I meant it for the authors, but didn't see email addresses
>> in the deck...
>>
>> --dmm
>>
>>
>> On Thu, Jun 30, 2016 at 10:44 AM, Sheng Jiang <jiangsheng@huawei.com>
>> wrote:
>>
>>> Hi, David,
>>>
>>>
>>>
>>> I am not sure your question is for me or for the speaker. Copied this
>>> email to the speaker, Panagiotis. If the question is on me, I don't really
>>> understand the point of your question. Could you please provide more
>>> context or explanation?
>>>
>>>
>>>
>>> Thanks,
>>>
>>>
>>>
>>> Sheng
>>>
>>>
>>> ------------------------------
>>> *From:* David Meyer [dmm@1-4-5.net]
>>> *Sent:* 30 June 2016 2:09
>>> *To:* Sheng Jiang
>>> *Cc:* nmlrg@irtf.org
>>> *Subject:* Re: [Nmlrg] links to each slide of presentations//RE: slides
>>> of NMLRG #3, June 27th, Athens
>>>
>>> This question is on this use case:
>>>
>>> Mobile network state characterization and prediction
>>>
>>> https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-4.pdf
>>>
>>> Isn't it the case that you will know if something is registered to the
>>> MBB, MMC, or MCC slice at device authentication time? If not what is the
>>> scenario in which you don't know?
>>>
>>> Thanks,
>>>
>>> --dmm
>>>
>>>
>>>
>>> On Tue, Jun 28, 2016 at 7:03 AM, Sheng Jiang <jiangsheng@huawei.com>
>>> wrote:
>>>
>>>> Oops... The proceeding page for interim meeting seems not be as
>>>> intelligent as the proceeding pages of IETF meetings. Our proceeding page
>>>> does not autonomically show the slides. I have sent an email to IETF
>>>> secretary to ask them to fix it. Meanwhile, in this email, here is the
>>>> links for each presentations:
>>>>
>>>> Slide Filename Edit Replace Delete
>>>> Chair Slides
>>>>
>>>> https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-0.pdf
>>>>
>>>> Introduction to Network Machine Learning & NMLRG
>>>>
>>>> https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-1.pdf
>>>>
>>>> Data Collection and Analysis At High Security Lab
>>>>
>>>> https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-2.pdf
>>>>
>>>> Use Cases of Applying Machine Learning Mechanism with Network Traffic
>>>>
>>>> https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-3.pdf
>>>>
>>>> Mobile network state characterization and prediction
>>>>
>>>> https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-4.pdf
>>>>
>>>> Learning how to route
>>>>
>>>> https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-5.pdf
>>>>
>>>> Regards,
>>>>
>>>> Sheng
>>>> ________________________________________
>>>> From: nmlrg [nmlrg-bounces@irtf.org] on behalf of Sheng Jiang [
>>>> jiangsheng@huawei.com]
>>>> Sent: 28 June 2016 21:03
>>>> To: nmlrg@irtf.org
>>>> Subject: [Nmlrg] slides of NMLRG #3, June 27th, Athens
>>>>
>>>> Hi, nmlrg,
>>>>
>>>> All slides that have been presented in our NMLRG #3 meeting, June 27th,
>>>> 2016, Athens, Greece, with EUCNC2016, have been uploaded. They can be
>>>> accessed through below link
>>>>
>>>>
>>>> https://www.ietf.org/proceedings/interim/2016/06/27/nmlrg/proceedings.html
>>>>
>>>> Best regards,
>>>>
>>>> Sheng
>>>> _______________________________________________
>>>> nmlrg mailing list
>>>> nmlrg@irtf.org
>>>> https://www.irtf.org/mailman/listinfo/nmlrg
>>>> _______________________________________________
>>>> nmlrg mailing list
>>>> nmlrg@irtf.org
>>>> https://www.irtf.org/mailman/listinfo/nmlrg
>>>>
>>>
>>>
>>
>

--94eb2c111ada98011205368331fd
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr">Thanks. Do you folks have a paper or other additional desc=
ription of your data and methods?<div><br></div><div>Thx,</div><div><br></d=
iv><div>--dmm</div><div><br></div></div><div class=3D"gmail_extra"><br><div=
 class=3D"gmail_quote">On Thu, Jun 30, 2016 at 11:28 AM, Panagiotis Demesti=
chas <span dir=3D"ltr">&lt;<a href=3D"mailto:pdemestichas@gmail.com" target=
=3D"_blank">pdemestichas@gmail.com</a>&gt;</span> wrote:<br><blockquote cla=
ss=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1px #ccc solid;pa=
dding-left:1ex"><div dir=3D"ltr">Dear all<br><div><br></div><div>just saw t=
he stream of emails.</div><div><br></div><div>The goal of the work is to le=
arn traffic patterns, and other associated aspects, e.g., in time and space=
. The goal is not to instantaneously identify the class of the stream; such=
 streams are used for fine tuning the learning.</div><div><br></div><div><d=
iv>Voula Vassaki (did the presentation) will complement.</div></div><div><b=
r></div><div>Best regards</div><span class=3D"HOEnZb"><font color=3D"#88888=
8"><div>Panagiotis</div><div><br></div><div><br></div></font></span></div><=
div class=3D"HOEnZb"><div class=3D"h5"><div class=3D"gmail_extra"><br><div =
class=3D"gmail_quote">On Thu, Jun 30, 2016 at 8:58 PM, David Meyer <span di=
r=3D"ltr">&lt;<a href=3D"mailto:dmm@1-4-5.net" target=3D"_blank">dmm@1-4-5.=
net</a>&gt;</span> wrote:<br><blockquote class=3D"gmail_quote" style=3D"mar=
gin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir=3D"ltr=
">Thanks Sheng, I meant it for the authors, but didn&#39;t see email addres=
ses in the deck...<div><br></div><div>--dmm</div><div><br><div class=3D"gma=
il_extra"><br><div class=3D"gmail_quote">On Thu, Jun 30, 2016 at 10:44 AM, =
Sheng Jiang <span dir=3D"ltr">&lt;<a href=3D"mailto:jiangsheng@huawei.com" =
target=3D"_blank">jiangsheng@huawei.com</a>&gt;</span> wrote:<br><blockquot=
e class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1px #ccc sol=
id;padding-left:1ex">




<div>
<div style=3D"FONT-FAMILY:Tahoma;DIRECTION:ltr;COLOR:#000000;FONT-SIZE:10pt=
">
<p>Hi, David,</p>
<p>=C2=A0</p>
<p>I am not sure your question is for me or for the speaker. Copied this em=
ail to the speaker, Panagiotis. If the question is on me, I don&#39;t reall=
y understand the point of your question. Could you please provide more cont=
ext or explanation?</p>
<p>=C2=A0</p>
<p>Thanks,</p>
<p>=C2=A0</p>
<p>Sheng</p>
<p>=C2=A0</p>
<div style=3D"FONT-FAMILY:Times New Roman;COLOR:#000000;FONT-SIZE:16px">
<hr>
<div style=3D"DIRECTION:ltr"><font color=3D"#000000" size=3D"2" face=3D"Tah=
oma"><b>From:</b> David Meyer [<a href=3D"mailto:dmm@1-4-5.net" target=3D"_=
blank">dmm@1-4-5.net</a>]<br>
<b>Sent:</b> 30 June 2016 2:09<span><br>
<b>To:</b> Sheng Jiang<br>
<b>Cc:</b> <a href=3D"mailto:nmlrg@irtf.org" target=3D"_blank">nmlrg@irtf.o=
rg</a><br>
</span><b>Subject:</b> Re: [Nmlrg] links to each slide of presentations//RE=
: slides of NMLRG #3, June 27th, Athens<br>
</font><br>
</div><div><div>
<div></div>
<div>
<div dir=3D"ltr">This question is on this use case:
<div><br>
<div>
<div>Mobile network state characterization and prediction</div>
<div><a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slid=
es/slides-interim-2016-nmlrg-01-4.pdf" target=3D"_blank">https://www.ietf.o=
rg/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-4.=
pdf</a></div>
</div>
<div><br>
</div>
<div>Isn&#39;t it the case that you will know if something is registered to=
 the MBB, MMC, or MCC slice at device authentication time? If not what is t=
he scenario in which you don&#39;t know?</div>
<div><br>
</div>
<div>Thanks,</div>
<div><br>
</div>
<div>--dmm</div>
<div><br>
</div>
<div><br>
<div class=3D"gmail_extra"><br>
<div class=3D"gmail_quote">On Tue, Jun 28, 2016 at 7:03 AM, Sheng Jiang <sp=
an dir=3D"ltr">
&lt;<a href=3D"mailto:jiangsheng@huawei.com" target=3D"_blank">jiangsheng@h=
uawei.com</a>&gt;</span> wrote:<br>
<blockquote style=3D"BORDER-LEFT:rgb(204,204,204) 1px solid;MARGIN:0px 0px =
0px 0.8ex;PADDING-LEFT:1ex" class=3D"gmail_quote">
Oops... The proceeding page for interim meeting seems not be as intelligent=
 as the proceeding pages of IETF meetings. Our proceeding page does not aut=
onomically show the slides. I have sent an email to IETF secretary to ask t=
hem to fix it. Meanwhile, in this
 email, here is the links for each presentations:<br>
<br>
Slide Filename Edit Replace Delete<br>
Chair Slides<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-0.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-0.pdf</a><br>
<br>
Introduction to Network Machine Learning &amp; NMLRG<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-1.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-1.pdf</a><br>
<br>
Data Collection and Analysis At High Security Lab<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-2.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-2.pdf</a><br>
<br>
Use Cases of Applying Machine Learning Mechanism with Network Traffic<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-3.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-3.pdf</a><br>
<br>
Mobile network state characterization and prediction<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-4.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-4.pdf</a><br>
<br>
Learning how to route<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-5.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-5.pdf</a><br>
<br>
Regards,<br>
<br>
Sheng<br>
________________________________________<br>
From: nmlrg [<a href=3D"mailto:nmlrg-bounces@irtf.org" target=3D"_blank">nm=
lrg-bounces@irtf.org</a>] on behalf of Sheng Jiang [<a href=3D"mailto:jiang=
sheng@huawei.com" target=3D"_blank">jiangsheng@huawei.com</a>]<br>
Sent: 28 June 2016 21:03<br>
To: <a href=3D"mailto:nmlrg@irtf.org" target=3D"_blank">nmlrg@irtf.org</a><=
br>
Subject: [Nmlrg] slides of NMLRG #3, June 27th, Athens<br>
<br>
Hi, nmlrg,<br>
<br>
All slides that have been presented in our NMLRG #3 meeting, June 27th, 201=
6, Athens, Greece, with EUCNC2016, have been uploaded. They can be accessed=
 through below link<br>
<br>
<a href=3D"https://www.ietf.org/proceedings/interim/2016/06/27/nmlrg/procee=
dings.html" rel=3D"noreferrer" target=3D"_blank">https://www.ietf.org/proce=
edings/interim/2016/06/27/nmlrg/proceedings.html</a><br>
<br>
Best regards,<br>
<br>
Sheng<br>
_______________________________________________<br>
nmlrg mailing list<br>
<a href=3D"mailto:nmlrg@irtf.org" target=3D"_blank">nmlrg@irtf.org</a><br>
<a href=3D"https://www.irtf.org/mailman/listinfo/nmlrg" rel=3D"noreferrer" =
target=3D"_blank">https://www.irtf.org/mailman/listinfo/nmlrg</a><br>
_______________________________________________<br>
nmlrg mailing list<br>
<a href=3D"mailto:nmlrg@irtf.org" target=3D"_blank">nmlrg@irtf.org</a><br>
<a href=3D"https://www.irtf.org/mailman/listinfo/nmlrg" rel=3D"noreferrer" =
target=3D"_blank">https://www.irtf.org/mailman/listinfo/nmlrg</a><br>
</blockquote>
</div>
<br>
</div>
</div>
</div>
</div>
</div>
</div></div></div>
</div>
</div>

</blockquote></div><br></div></div></div>
</blockquote></div><br></div>
</div></div></blockquote></div><br></div>

--94eb2c111ada98011205368331fd--


From nobody Thu Jun 30 12:16:41 2016
Return-Path: <leo.liubing@huawei.com>
X-Original-To: nmlrg@ietfa.amsl.com
Delivered-To: nmlrg@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 1812F12DB6D for <nmlrg@ietfa.amsl.com>; Thu, 30 Jun 2016 12:16:40 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -5.646
X-Spam-Level: 
X-Spam-Status: No, score=-5.646 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, HTML_MESSAGE=0.001, RCVD_IN_DNSWL_MED=-2.3, RCVD_IN_MSPIKE_H3=-0.01, RCVD_IN_MSPIKE_WL=-0.01, RP_MATCHES_RCVD=-1.426, SPF_PASS=-0.001] autolearn=ham autolearn_force=no
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id Uhp45UMrR8uv for <nmlrg@ietfa.amsl.com>; Thu, 30 Jun 2016 12:16:34 -0700 (PDT)
Received: from szxga03-in.huawei.com (szxga03-in.huawei.com [119.145.14.66]) (using TLSv1 with cipher RC4-SHA (128/128 bits)) (No client certificate requested) by ietfa.amsl.com (Postfix) with ESMTPS id E2DD112DB61 for <nmlrg@irtf.org>; Thu, 30 Jun 2016 12:16:32 -0700 (PDT)
Received: from 172.24.1.36 (EHLO NKGEML413-HUB.china.huawei.com) ([172.24.1.36]) by szxrg03-dlp.huawei.com (MOS 4.4.3-GA FastPath queued) with ESMTP id CED90411; Fri, 01 Jul 2016 03:13:14 +0800 (CST)
Received: from NKGEML514-MBX.china.huawei.com ([fe80::40a8:f0d:c0f3:2ca5]) by NKGEML413-HUB.china.huawei.com ([10.98.56.74]) with mapi id 14.03.0235.001; Fri, 1 Jul 2016 03:13:10 +0800
From: "Liubing (Leo)" <leo.liubing@huawei.com>
To: David Meyer <dmm@1-4-5.net>
Thread-Topic: [Nmlrg] links to each slide of presentations//RE: slides of NMLRG #3, June 27th, Athens
Thread-Index: AQHR0UXZfny1k5v9oUOv37+HaylYt5/+e2IAgAGH4UyAAAvZgIACTvMQ
Date: Thu, 30 Jun 2016 19:13:09 +0000
Message-ID: <8AE0F17B87264D4CAC7DE0AA6C406F45C2DB62D6@nkgeml514-mbx.china.huawei.com>
References: <5D36713D8A4E7348A7E10DF7437A4B927CA893E9@NKGEML515-MBX.china.huawei.com> <CAHiKxWhfdukjRVnSKnbLwMwamiqZYoCD350BoNqO4njqkQoU1g@mail.gmail.com> <5D36713D8A4E7348A7E10DF7437A4B927CA897D4@NKGEML515-MBX.china.huawei.com> <CAHiKxWhMPjKY24zucg-UND0ARWHbjnOyYv=Tia-yk-Vq6Ta_SA@mail.gmail.com>
In-Reply-To: <CAHiKxWhMPjKY24zucg-UND0ARWHbjnOyYv=Tia-yk-Vq6Ta_SA@mail.gmail.com>
Accept-Language: en-US, zh-CN
Content-Language: zh-CN
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
x-originating-ip: [10.47.78.4]
Content-Type: multipart/alternative; boundary="_000_8AE0F17B87264D4CAC7DE0AA6C406F45C2DB62D6nkgeml514mbxchi_"
MIME-Version: 1.0
X-CFilter-Loop: Reflected
X-Mirapoint-Virus-RAPID-Raw: score=unknown(0), refid=str=0001.0A090204.57756F4B.00F0, ss=1, re=0.000, recu=0.000, reip=0.000,  cl=1, cld=1, fgs=0, ip=0.0.0.0, so=2013-05-26 15:14:31, dmn=2013-03-21 17:37:32
X-Mirapoint-Loop-Id: a75f6fa6d7c1294bb5b6a58d5dcf8817
Archived-At: <https://mailarchive.ietf.org/arch/msg/nmlrg/mymYAJ3Vyg7IgEfdx7gYGzy-_gs>
Cc: "nmlrg@irtf.org" <nmlrg@irtf.org>
Subject: Re: [Nmlrg] links to each slide of presentations//RE: slides of NMLRG #3, June 27th, Athens
X-BeenThere: nmlrg@irtf.org
X-Mailman-Version: 2.1.17
Precedence: list
List-Id: Network Machine Learning Research Group <nmlrg.irtf.org>
List-Unsubscribe: <https://www.irtf.org/mailman/options/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=unsubscribe>
List-Archive: <https://mailarchive.ietf.org/arch/browse/nmlrg/>
List-Post: <mailto:nmlrg@irtf.org>
List-Help: <mailto:nmlrg-request@irtf.org?subject=help>
List-Subscribe: <https://www.irtf.org/mailman/listinfo/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=subscribe>
X-List-Received-Date: Thu, 30 Jun 2016 19:16:40 -0000

--_000_8AE0F17B87264D4CAC7DE0AA6C406F45C2DB62D6nkgeml514mbxchi_
Content-Type: text/plain; charset="utf-8"
Content-Transfer-Encoding: base64
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--_000_8AE0F17B87264D4CAC7DE0AA6C406F45C2DB62D6nkgeml514mbxchi_
Content-Type: text/html; charset="utf-8"
Content-Transfer-Encoding: base64
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--_000_8AE0F17B87264D4CAC7DE0AA6C406F45C2DB62D6nkgeml514mbxchi_--


From nobody Thu Jun 30 13:20:13 2016
Return-Path: <jiangsheng@huawei.com>
X-Original-To: nmlrg@ietfa.amsl.com
Delivered-To: nmlrg@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 4A57812D59F for <nmlrg@ietfa.amsl.com>; Thu, 30 Jun 2016 13:20:11 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -5.646
X-Spam-Level: 
X-Spam-Status: No, score=-5.646 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, HTML_MESSAGE=0.001, RCVD_IN_DNSWL_MED=-2.3, RCVD_IN_MSPIKE_H4=-0.01, RCVD_IN_MSPIKE_WL=-0.01, RP_MATCHES_RCVD=-1.426, SPF_PASS=-0.001] autolearn=ham autolearn_force=no
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 2oNu0CZk2BJQ for <nmlrg@ietfa.amsl.com>; Thu, 30 Jun 2016 13:20:06 -0700 (PDT)
Received: from szxga02-in.huawei.com (szxga02-in.huawei.com [119.145.14.65]) (using TLSv1 with cipher RC4-SHA (128/128 bits)) (No client certificate requested) by ietfa.amsl.com (Postfix) with ESMTPS id 9105412B01C for <nmlrg@irtf.org>; Thu, 30 Jun 2016 13:20:05 -0700 (PDT)
Received: from 172.24.1.137 (EHLO NKGEML413-HUB.china.huawei.com) ([172.24.1.137]) by szxrg02-dlp.huawei.com (MOS 4.3.7-GA FastPath queued) with ESMTP id DJP89542; Fri, 01 Jul 2016 04:16:47 +0800 (CST)
Received: from NKGEML515-MBX.china.huawei.com ([fe80::a54a:89d2:c471:ff]) by NKGEML413-HUB.china.huawei.com ([10.98.56.74]) with mapi id 14.03.0235.001; Fri, 1 Jul 2016 04:16:42 +0800
From: Sheng Jiang <jiangsheng@huawei.com>
To: David Meyer <dmm@1-4-5.net>
Thread-Topic: [Nmlrg] links to each slide of presentations//RE: slides of NMLRG #3, June 27th, Athens
Thread-Index: AQHR0UXZfny1k5v9oUOv37+HaylYt5/+e2IAgAGH4UyAAAvZgIACSbUQ
Date: Thu, 30 Jun 2016 20:16:42 +0000
Message-ID: <5D36713D8A4E7348A7E10DF7437A4B927CA89FB4@NKGEML515-MBX.china.huawei.com>
References: <5D36713D8A4E7348A7E10DF7437A4B927CA893E9@NKGEML515-MBX.china.huawei.com> <CAHiKxWhfdukjRVnSKnbLwMwamiqZYoCD350BoNqO4njqkQoU1g@mail.gmail.com> <5D36713D8A4E7348A7E10DF7437A4B927CA897D4@NKGEML515-MBX.china.huawei.com>, <CAHiKxWhMPjKY24zucg-UND0ARWHbjnOyYv=Tia-yk-Vq6Ta_SA@mail.gmail.com>
In-Reply-To: <CAHiKxWhMPjKY24zucg-UND0ARWHbjnOyYv=Tia-yk-Vq6Ta_SA@mail.gmail.com>
Accept-Language: en-GB, zh-CN, en-US
Content-Language: en-GB
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
x-originating-ip: [10.47.64.200]
Content-Type: multipart/alternative; boundary="_000_5D36713D8A4E7348A7E10DF7437A4B927CA89FB4NKGEML515MBXchi_"
MIME-Version: 1.0
X-CFilter-Loop: Reflected
X-Mirapoint-Virus-RAPID-Raw: score=unknown(0), refid=str=0001.0A090206.57757E30.0065, ss=1, re=0.000, recu=0.000, reip=0.000,  cl=1, cld=1, fgs=0, ip=0.0.0.0, so=2013-06-18 04:22:30, dmn=2013-03-21 17:37:32
X-Mirapoint-Loop-Id: d3da655d6566e22179791efb37db52bc
Archived-At: <https://mailarchive.ietf.org/arch/msg/nmlrg/U4_31jP7vy0J2Mf4u9vBHYyA2jU>
Cc: "nmlrg@irtf.org" <nmlrg@irtf.org>
Subject: Re: [Nmlrg] links to each slide of presentations//RE: slides of NMLRG #3, June 27th, Athens
X-BeenThere: nmlrg@irtf.org
X-Mailman-Version: 2.1.17
Precedence: list
List-Id: Network Machine Learning Research Group <nmlrg.irtf.org>
List-Unsubscribe: <https://www.irtf.org/mailman/options/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=unsubscribe>
List-Archive: <https://mailarchive.ietf.org/arch/browse/nmlrg/>
List-Post: <mailto:nmlrg@irtf.org>
List-Help: <mailto:nmlrg-request@irtf.org?subject=help>
List-Subscribe: <https://www.irtf.org/mailman/listinfo/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=subscribe>
X-List-Received-Date: Thu, 30 Jun 2016 20:20:11 -0000

--_000_5D36713D8A4E7348A7E10DF7437A4B927CA89FB4NKGEML515MBXchi_
Content-Type: text/plain; charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable

Hi, David,



Please see my replies with <Sheng/> in the lines. Sorry for being slow resp=
onse due to my traveling.



However, upto now, I have serveral concerns. Firstly, before we could discu=
ss such dataset, we need to narrow our target scenarios to a few very speci=
fic use cases, giving that there are so many various scenarios in the netwo=
rk area and they are so different from each other and require different dat=
aset.

I'm hoping that not every network is not a one off. Rather, what we seek is=
 to learn what are the fundamental properties of networks that can be the b=
asis of reasoning about networks do. For example, we should be able to lear=
n something from work done on transfer learning [0] and representation lear=
ning [1]. In the image recognition space there are even startups that provi=
de pretrained AlexNet, GooLeNet, etc trained on ImageNet (and others) that =
you can further train/customize for your particular application. In any eve=
nt, if every network requires something totally different then we haven't y=
et understood what is fundamental to the structure of our networks.

<Sheng>If I understand you correctly, you meant we should do some generic l=
earning in order to better understand the network fundamentally, and the re=
sult should be helpful for solving these specific tasks in the networks. It=
 is different to what we have done - working on applying ML to solve specif=
ic network tasks. I would agree this proposed research direction are import=
ant and  I am very interesting in such direction although  it looks ambitio=
ns and no guarantee for success so far. </Sheng>


Secondly, as we did talk with operators, the real data is the important pro=
perties for them. So, it would be almost impossible for them to share them.=
 It leaves two possibilities on how to get dataset: a) generating dataset u=
sing simulation;

This likely won't work. For example, if you do this then try to train a DNN=
 on that data set the NN will learn how the simulator works. This is unlike=
ly to be the outcome that we'd like (we already know how the simulator work=
s).

 b) try to get some partial dataset from operators that have removed their =
sensenble data. Both way are problemstic, in the risk that a lot of efforts=
 may waste in wrong areas.

Agree that this is problematic.

<Sheng>Here, we seems reach the agreement on that we do NOT have a good way=
 to obtain the useful dataset. It is not good at all. Does that mean we hav=
e to wait for operators to work on themselves alone? Any suggestion on how =
we could get quality real dataset?</Sheng>
Thirdly, the availability of data may very different in different network e=
nvironments. It depends on the measurement/congnitive functions that have b=
een deployed or implemented.

What is a "cognitive function"?

<Sheng>Here, I use the "cognitive function" to represent that device or net=
work function that collects the existing status or configuration data, agai=
nst measurement.</Sheng>

In any event, you have outlined just a few of the problems with creating "s=
tandardized" data sets for our use. I have quoted standardized because I'm =
beginning to think its the wrong word (invokes "standards" e.g., IETF, <X>.=
.., and that is not what is meant here).  Not sure what the right term is b=
ut  "standardized" seems to be the wrong word here.




Although having the above concerns, I  think we should: a) research on cert=
ain dataset. It would be very useful to prove the machine learning could pe=
rform better or provide more flexibility and adaptibility;

Not sure what you mean here. By "certain dataset" you mean what?

<Sheng>We did discuss a potential standardized dataset, didn't we?</Sheng>


b) continue our discuss on various network use cases that could benefit fro=
m applying machine learning, in order to serve wider network communities.

Use cases are good, however machine learning isn't something we can just ap=
ply to a use case because we've defined the use case. There are really at l=
east three parts of this that need to be considered as part of a solution: =
(i). use case (as you mention), (ii). algorithm(s) that are going to provid=
e the regression/classification/ desired in (i), and (iii). data sets.

(ii). requires understanding of the algorithms that can be applied to produ=
ce the inferences (regressions, ...)  we are interested in. What algorithms=
 we can use depend on the data we can get (iii).

<Sheng>What we have done up to now is, in a specific use case, to design an=
 algorithm or a combination of algorithm on chosen dataset (this may be dyn=
amic data from the network environment) to complete a specific task.</Sheng=
>

That said, one of the things we've learned over the past decade or so is th=
at learning itself is a very general process. We see this in how powerful s=
omething as simple as SGD/backprop is (see e.g., [2]). We can also use this=
 to our advantage and re-purpose algorithms originally developed for other =
tasks. A simple example of this is LDA (see [3,4]). Here the original idea =
was to model topics in a corpus of documents, where the topics themselves (=
along with the per-document topic proportions) were the latent/hidden varia=
bles. One way to re-purpose LDA for applications like ours (essentially cla=
ssification/anomaly detection tasks) is to view each to/from pair of IP add=
resses as a "document" (in the language of LDA) and then words are construc=
ted from features of the to/from traffic (e.g., to ports, etc). See [5] for=
 a decent example of how you might think about this kind of "statistical cl=
ustering" using the LDA Probabilistic Graphical Model approach. So there is=
 hope, but there is a lot of work left to be done on both the theory and pr=
actice sides of this (which as I mentioned earlier are, at least currently,=
 essentially the same thing).

<Sheng>Agree on that learning and learning algorithm is general. But what w=
e did was only use it in specific way to solve specific tasks/use cases. As=
 I agreed earlier, generlized learning on network could be an important dir=
ection and I am interesting in. Let's see how far we could go. On another s=
ide, I have to emphasis that the general learning must be helpful on solvin=
g specific tasks, in other words, specific tasks should be earlier to solve=
 from the result of general learning. Otherwise, the general learning is no=
t mearningful.</Sheng>


Another important point I would like to discuss is that you mentioned "our =
goal is to provide accurate, repeatable, and explainable results." Ideally,=
 we do want this. It is our traditional way for logic-based programming.

What is "logic-based programming"?  Do you mean logic programming, e.g., Pr=
olog? If so I understand logic programming (I did a lot of work on logic pr=
ogramming back in the late 80s/early 90s, see http://dl.acm.org/citation.cf=
m?id=3D87991). But even so, I don't understand the connection, so can you b=
e more explicit?

<Sheng>Here, "logic-based programming" I referred to all traditional progra=
mming that realize human pre-designed logic into computing programs. It is =
comparing to ML-based programming.</Sheng>


However, I am not sure that is a doable target with machine learning. Many =
of machine learning algorithms are based on statistics theory, which may no=
t be that much accurate.

Sorry, I don't understand this. Things like Bayesian inference (e.g., LDA) =
are inherently about quantifying uncertainty. In any event can you be more =
specific? A learning algorithm can "accurately" approximate something like =
a posterior distribution, its just that the output is just that, a posterio=
r distribution. So there is a difference between "accuracy" (which isn't re=
ally defined here) and the stochastic nature of the results we get from ML.

OTOH, if you mean accuracy of our estimated distributions then there are ob=
viously ways this can go wrong. For example, even an algorithm as simple as=
 k-means can go wrong since the optimization objective (distortion function=
) is non-convex, so it is sensitive to initial conditions (in particular, w=
here you start; if you are unlucky you can get stuck in a local minimum  du=
e to the non-convex nature of the objective). See e.g., [8]. BTW,  k-means =
was called out in one of the decks below but as I mentioned, not much discu=
ssion of the important topics we are discussing here there. There is also a=
pproximate inference (for example, MCMC or variational approaches), but the=
se approaches also have provably quantitative error bounds (http://www.1-4-=
5.net/~dmm/ml/ps.pdf talks a bit about how variational approaches, essentia=
l trading inference for optimization where the inference (integral) is intr=
actable, find a tight lower-bound using Jensen's inequality.

<Sheng>The reason that statistics-based ML algorithms can work is that they=
 chose to focus on only high probability and igrone corner cases. But when =
mearningful corner cases happen or we want to target the corner cases, mach=
ine learning may not accurately work. Again, "accuracy" itself does not pro=
perly defined. So, we may talk about very different things. On other side, =
I also agree that accurate is not important topic here. ML can reach a high=
-level accurate, which are suffient for majority cases. But ML may never re=
ach full accurate while traditional program can 100% certain the result on =
giving circumstance.</Sheng>


Some of machine learning algorithms are black box or have very complexicate=
d internal computing process. So, the resulf of these algorithms may be har=
d to be explained.

This is definitely true to some extent today with DNNs, but that is changin=
g rapidly. Results can be repeatable but explaining exactly what units (art=
ificial neurons) far from the input are doing can be difficult. But this to=
o is an area of active research and development (see [6] from last year's I=
CML, or the work I pointed to on GANs in my last note as progress in this a=
rea). Other approaches such probabilistic graphical models or signal proces=
sing approaches are perhaps more easily explained, but I don't think this i=
s the level of explanation that is going to be useful to people deploying a=
nd operating networks, so we'll have to find different ways of explaining w=
hat things like DNNs to in a way that is consumable by operators, etc.

<Sheng>Sharing the same opinion here. Every we could really explain how DNN=
s works in a very complicated way. Our customers would not have enough pati=
ent to listen. They would still not trust ML. We, actually the whole AI/ML =
society, needs to find other way to explain and prove things like DNN are t=
rustable. Only then, they can be use for more serious or autonomic tasks.</=
Sheng>

All this by way of saying that we are gaining better understanding how DNNs=
 work every day (literally) and as a result  I'm optimistic that we can bui=
ld technology that will be accurate (again, stochastic results), repeatable=
 and explainable.

<Sheng>I hope I can share you optimistic. Actually, I do believe we could d=
o so one day. But I am not sure how long it will take. </Sheng>

I'll just note here that if you have a model with literally billions of par=
ameters being trained by SGD its not that surprising that humans have a har=
d time understanding what the computation is. As a concrete example, DNNs c=
an solve the so-called "selectivity-invariance" problem [7], but we humans =
don't really  know how to do, so this is another example.


Also the process of machine learning may be hard to be controlled or interv=
ened.

Again, not sure what you mean here. Please give a bit more detail.

<Sheng> Here, I actually refer to DNN too. We don't know how to intervened =
DNN in its running time.


Repeatable, yes, we have to have this feature to be widely adopted. However=
, repeatable may have to come with some level of toleration for the inaccur=
ate.

All of this is inherently probabilistic. A DNN, for example, can output the=
 posterior distribution over some hidden variables of interest accurately. =
LDA finds the latent "topics" and word distributions, both of which are mul=
tinomals so again we're essentially getting a posterior over the hidden var=
iables.  Even k-means is  just the EM (Expectation Maximization) algorithm =
applied to a particular Naive Bayes model. The same can be said of almost e=
very learning algorithm you might mention. So "inaccurate" isn't the correc=
t word here (unless you mean whatever you trained didn't predict whatever y=
ou wanted to predict, in which case something else is wrong). So the DNN gi=
ves you as stochastic result, which may be different than what many network=
 people are used to. But its not inaccurate, its just stochastic.

<Sheng>Inaccurate or stochastic, they are not what many network people can =
accept. In their eyes, network is definite, every parameter is definite. Ho=
wever, actually, if you could analyze every network parameter, the value ma=
y be definite, but actually they all have toleration for some level of inac=
curate as long as their design purposes could meet. It is the fundemental t=
hat we could use ML to solve network task.

Thanks so much. Regards,

Sheng </Sheng>



Thx,

--dmm



[0] http://people.csail.mit.edu/mcollins/papers/CVPR_2008.pdf
[1]  https://arxiv.org/abs/1206.5538
[2] http://deliprao.com/archives/153
[3] http://www.1-4-5.net/~dmm/ml/lda_intro.pdf
[4] http://www.jmlr.org/papers/volume3/blei03a/blei03a.pdf
[5] http://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=3D5949386
[6] http://yosinski.com/media/papers/Yosinski__2015__ICML_DL__Understanding=
_Neural_Networks_Through_Deep_Visualization__.pdf
[7] https://cbmm.mit.edu/sites/default/files/publications/Anselmi_Invarianc=
e_CBMM_memo_29.pdf
[8] http://cs229.stanford.edu/notes/cs229-notes7a.pdf

Best regards,



Sheng



________________________________

From: David Meyer [dmm@1-4-5.net<mailto:dmm@1-4-5.net>]
Sent: 28 June 2016 23:30http://deliprao.com/archives/153
To: Sheng Jiang
Cc: nmlrg@irtf.org<mailto:nmlrg@irtf.org>
Subject: Re: [Nmlrg] links to each slide of presentations//RE: slides of NM=
LRG #3, June 27th, Athens

Sheng,

Thanks for the pointers. One thing I notice is that we don't have much on w=
hat the characteristics of network data might be, and as such, what kinds o=
f existing learning algorithms might be suitable and where additional devel=
opment might be required. For example, is the collected data IID? Is it tim=
e series? Is the underlying distribution stationary? If not, how does that =
constrain algorithms we might use or develop? For example, how does a given=
 algorithm deal with concept drift/internal covariate shift (in the case of=
 DNNs; see e.g., batch normalization [0]). There are many other such questi=
ons, such as is the data categorical (e.g., ports, IP addresses) or is it c=
ontinuous/discrete (e.g., counters). And if the data in question is categor=
ical, what is the cardinality of the categories (this will inform how such =
data can be encoded); in the case of IP addresses we can't really one-hot e=
ncode addresses because their cardinality is too large (2^32 or 2^128); thi=
s has implications for how we build classifiers (in particular, for softmax=
 layers in DNNs of various kinds).

Related to the above is the question of features? What are good features fo=
r networking? Where do they come from? Are they domain specific? Can we lea=
rn features the way a DNNs does in the network space? Can we use autoencode=
rs to discover such features? Or can we use GANs to train DNNs for network =
classification tasks in an unsupervised manner? Are there other, non-ad-hoc=
 (well founded) methods we can use, or is every use case a one-off (one wou=
ld hope not).

We can carry the same kinds of analyzes to the algorithms applied. For exam=
ple, while something k-means is an effective way to get a feeling for how c=
ontinuous/discrete hangs together, if our data is categorical statistical c=
lustering approaches such as LDA might provide a more well-founded approach=
 (of course, as with most Bayesian techniques is the question of approximat=
e inference arises since in most interesting cases the integral that we nee=
d to solve, namely the marginal probability data isn't tractable so we need=
 to resort to MCMC or more likely variational inference). And what about th=
e use of SGD/batch normalization etc with DNNs, and perhaps more importantl=
y, can we use network data to train DNN policy networks for reinforcement l=
earning like we saw in deep Q-learning and AlphaGo?

These comments are all by way of saying that we don't have a solid theoreti=
cal understanding (yet) of how techniques that have been so successful in o=
ther domains (e.g., DNNs for perceptual tasks) generalize to networking use=
 cases. We will need this understanding if our goal is to provide accurate,=
 repeatable, and explainable results.

In order to accomplish all of this we need, as I have been saying , not onl=
y a good understanding of how these algorithms work but also standardized d=
ata sets and associated benchmarks so we can tell if we are making progress=
 (or even if our techniques work). Analogies here include MNIST and ImageNe=
t and their associated benchmarks, among others. As mentioned  standardized=
 data sets are key to making progress in the ML for networking space (other=
wise how do you know your technique works and/or improves on another techni=
ques?). One might assume that these data sets would need to be labeled (as =
supervised learning is where most of the progress is being made these days)=
, but not necessarily; Generative Adversarial Networks (GANs) have emerged =
as a new way to train DNNs in an unsupervised manner (this is moving very r=
apidly; see e.g., https://openai.com/blog/generative-models/).

The summary here is that the "distance" between theory and practice in ML i=
s effectively zero right now due to the incredible rate of progress in the =
field; this means we need  to understand both sides of the theory/practice =
coin in order to be effective. None of the slide decks provide much backgro=
und on what the proposed algorithms are, how they work, or why they should =
be expected to work on network data.

Finally, if you are interested in LDA or other algorithms there are a few s=
hort explanatory pieces I have written for my team on http://www.1-4-5.net/=
~dmm/ml (works in progress).

Thanks,

Dave

[0] https://arxiv.org/pdf/1502.03167.pdf

On Tue, Jun 28, 2016 at 7:03 AM, Sheng Jiang <jiangsheng@huawei.com<mailto:=
jiangsheng@huawei.com>> wrote:
Oops... The proceeding page for interim meeting seems not be as intelligent=
 as the proceeding pages of IETF meetings. Our proceeding page does not aut=
onomically show the slides. I have sent an email to IETF secretary to ask t=
hem to fix it. Meanwhile, in this email, here is the links for each present=
ations:

Slide Filename Edit Replace Delete
Chair Slides
https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interi=
m-2016-nmlrg-01-0.pdf

Introduction to Network Machine Learning & NMLRG
https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interi=
m-2016-nmlrg-01-1.pdf

Data Collection and Analysis At High Security Lab
https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interi=
m-2016-nmlrg-01-2.pdf

Use Cases of Applying Machine Learning Mechanism with Network Traffic
https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interi=
m-2016-nmlrg-01-3.pdf

Mobile network state characterization and prediction
https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interi=
m-2016-nmlrg-01-4.pdf

Learning how to route
https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interi=
m-2016-nmlrg-01-5.pdf

Regards,

Sheng
________________________________________
From: nmlrg [nmlrg-bounces@irtf.org<mailto:nmlrg-bounces@irtf.org>] on beha=
lf of Sheng Jiang [jiangsheng@huawei.com<mailto:jiangsheng@huawei.com>]
Sent: 28 June 2016 21:03
To: nmlrg@irtf.org<mailto:nmlrg@irtf.org>
Subject: [Nmlrg] slides of NMLRG #3, June 27th, Athens

Hi, nmlrg,

All slides that have been presented in our NMLRG #3 meeting, June 27th, 201=
6, Athens, Greece, with EUCNC2016, have been uploaded. They can be accessed=
 through below link

https://www.ietf.org/proceedings/interim/2016/06/27/nmlrg/proceedings.html

Best regards,

Sheng
_______________________________________________
nmlrg mailing list
nmlrg@irtf.org<mailto:nmlrg@irtf.org>
https://www.irtf.org/mailman/listinfo/nmlrg
_______________________________________________
nmlrg mailing list
nmlrg@irtf.org<mailto:nmlrg@irtf.org>
https://www.irtf.org/mailman/listinfo/nmlrg



--_000_5D36713D8A4E7348A7E10DF7437A4B927CA89FB4NKGEML515MBXchi_
Content-Type: text/html; charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable

<html dir=3D"ltr">
<head>
<meta http-equiv=3D"Content-Type" content=3D"text/html; charset=3Diso-8859-=
1">
<style id=3D"owaParaStyle">P {
	MARGIN-TOP: 0px; MARGIN-BOTTOM: 0px
}
</style>
</head>
<body fPStyle=3D"1" ocsi=3D"0">
<div style=3D"direction: ltr;font-family: Tahoma;color: #000000;font-size: =
10pt;">
<p>Hi, David,</p>
<p>&nbsp;</p>
<p>Please see my replies with &lt;Sheng/&gt; in the lines.&nbsp;Sorry for b=
eing slow response due to my traveling.</p>
<p>&nbsp;</p>
<div style=3D"FONT-FAMILY: Times New Roman; COLOR: #000000; FONT-SIZE: 16px=
">
<div>
<div dir=3D"ltr">
<div class=3D"gmail_extra">
<div class=3D"gmail_quote">
<blockquote style=3D"BORDER-LEFT: rgb(204,204,204) 1px solid; MARGIN: 0px 0=
px 0px 0.8ex; PADDING-LEFT: 1ex" class=3D"gmail_quote">
<div>
<div style=3D"FONT-FAMILY: Tahoma; DIRECTION: ltr; COLOR: rgb(0,0,0); FONT-=
SIZE: 10pt">
<p>However, upto now, I have serveral concerns. Firstly, before we could di=
scuss such dataset, we need to narrow our target scenarios to a few very&nb=
sp;specific use cases, giving that there are so many various scenarios in t=
he network area and they are so different
 from each other and require different dataset. </p>
</div>
</div>
</blockquote>
<div><br>
</div>
<div>I'm hoping that not every network is not a one off. Rather, what we se=
ek is to learn what are the fundamental properties of networks that can be =
the basis of reasoning about networks do. For example, we should be able to=
 learn something from work done
 on transfer learning [0] and representation learning [1]. In the image rec=
ognition space there are even startups that provide pretrained AlexNet, Goo=
LeNet, etc trained on ImageNet (and others) that you can further train/cust=
omize for your particular application.
 In any event, if every network requires something totally different then w=
e haven't yet understood what is fundamental to the structure of our networ=
ks.</div>
<div>&nbsp;</div>
<div>&lt;Sheng&gt;If I understand you correctly, you meant we should do som=
e generic learning in order to better understand the network fundamentally,=
 and the result should be helpful for solving these specific tasks in the n=
etworks. It is different to what we have
 done - working on applying ML to solve specific network tasks. I would&nbs=
p;agree this proposed research direction are important and&nbsp; I am very =
interesting in such direction although&nbsp; it looks ambitions and no guar=
antee for success so far. &lt;/Sheng&gt;</div>
<div>&nbsp;</div>
<blockquote style=3D"BORDER-LEFT: rgb(204,204,204) 1px solid; MARGIN: 0px 0=
px 0px 0.8ex; PADDING-LEFT: 1ex" class=3D"gmail_quote">
<div>
<div style=3D"FONT-FAMILY: Tahoma; DIRECTION: ltr; COLOR: rgb(0,0,0); FONT-=
SIZE: 10pt">
<p>Secondly, as we did talk with operators, the real data is the important =
properties for them. So, it would be almost impossible for them to share th=
em. It leaves two possibilities on how to get dataset: a) generating datase=
t using simulation;</p>
</div>
</div>
</blockquote>
<div><br>
</div>
<div>This likely won't work. For example, if you do this then try to train =
a DNN on that data set the NN will learn how the simulator works. This is u=
nlikely to be the outcome that we'd like (we already know how the simulator=
 works).</div>
<div><br>
</div>
<div>&nbsp;<span style=3D"FONT-FAMILY: Tahoma; COLOR: rgb(0,0,0); FONT-SIZE=
: 10pt">b) try to get some partial dataset from operators that have removed=
 their sensenble data. Both way are problemstic, in the risk that a lot of =
efforts may waste in wrong areas.&nbsp;</span></div>
<div><span style=3D"FONT-FAMILY: Tahoma; COLOR: rgb(0,0,0); FONT-SIZE: 10pt=
"><br>
</span></div>
<div><span style=3D"FONT-FAMILY: Tahoma; COLOR: rgb(0,0,0); FONT-SIZE: 10pt=
">Agree that this is problematic.&nbsp;</span></div>
<span style=3D"FONT-FAMILY: Tahoma; COLOR: rgb(0,0,0); FONT-SIZE: 10pt"></s=
pan></div>
<span style=3D"FONT-FAMILY: Tahoma; COLOR: rgb(0,0,0); FONT-SIZE: 10pt"></s=
pan></div>
<span style=3D"FONT-FAMILY: Tahoma; COLOR: rgb(0,0,0); FONT-SIZE: 10pt"></s=
pan></div>
<span style=3D"FONT-FAMILY: Tahoma; COLOR: rgb(0,0,0); FONT-SIZE: 10pt"></s=
pan></div>
<div><span style=3D"FONT-FAMILY: Tahoma; COLOR: rgb(0,0,0); FONT-SIZE: 10pt=
">
<div dir=3D"ltr" class=3D"gmail_quote">&nbsp;</div>
<div dir=3D"ltr" class=3D"gmail_quote">&lt;Sheng&gt;Here, we seems reach th=
e agreement on that we do NOT&nbsp;have a good way to obtain the useful dat=
aset. It is not good at all. Does that mean we have to wait for operators t=
o work on themselves alone? Any suggestion on how
 we could get quality real dataset?&lt;/Sheng&gt;<br>
</div>
</span>
<div dir=3D"ltr" class=3D"gmail_quote"><span style=3D"FONT-FAMILY: Tahoma; =
COLOR: rgb(0,0,0); FONT-SIZE: 10pt">Thirdly, the availability of data&nbsp;=
may very different in different network environments. It depends on the mea=
surement/congnitive functions that have been
 deployed or implemented. &nbsp;&nbsp;</span></div>
<div dir=3D"ltr" class=3D"gmail_quote"><span style=3D"FONT-FAMILY: Tahoma; =
COLOR: rgb(0,0,0); FONT-SIZE: 10pt"><br>
</span></div>
<div dir=3D"ltr" class=3D"gmail_quote"><span style=3D"FONT-FAMILY: Tahoma; =
COLOR: rgb(0,0,0); FONT-SIZE: 10pt">What is a &quot;cognitive function&quot=
;?</span></div>
<div dir=3D"ltr" class=3D"gmail_quote"><span style=3D"FONT-FAMILY: Tahoma; =
COLOR: rgb(0,0,0); FONT-SIZE: 10pt"></span>&nbsp;</div>
<div dir=3D"ltr" class=3D"gmail_quote"><span style=3D"FONT-FAMILY: Tahoma; =
COLOR: rgb(0,0,0); FONT-SIZE: 10pt">&lt;Sheng&gt;Here, I use the &quot;cogn=
itive function&quot; to represent that device or network function that&nbsp=
;collects the existing status or&nbsp;configuration data, against
 measurement.&lt;/Sheng&gt;</span></div>
<div dir=3D"ltr" class=3D"gmail_quote"><span style=3D"FONT-FAMILY: Tahoma; =
COLOR: rgb(0,0,0); FONT-SIZE: 10pt"><br>
</span></div>
<div dir=3D"ltr" class=3D"gmail_quote"><span style=3D"FONT-FAMILY: Tahoma; =
COLOR: rgb(0,0,0); FONT-SIZE: 10pt">In any event, you have outlined just a =
few of the problems with creating &quot;standardized&quot; data sets for ou=
r use. I have quoted standardized because I'm beginning
 to think its the wrong word (invokes &quot;standards&quot; e.g., IETF, &lt=
;X&gt;..., and that is not what is meant here).&nbsp; Not sure what the rig=
ht term is but &nbsp;&quot;standardized&quot; seems to be the wrong word he=
re.</span></div>
<div dir=3D"ltr" class=3D"gmail_quote"><span style=3D"FONT-FAMILY: Tahoma; =
COLOR: rgb(0,0,0); FONT-SIZE: 10pt"><br>
</span></div>
<blockquote style=3D"BORDER-LEFT: rgb(204,204,204) 1px solid; MARGIN: 0px 0=
px 0px 0.8ex; PADDING-LEFT: 1ex" dir=3D"ltr" class=3D"gmail_quote">
<div style=3D"FONT-FAMILY: Tahoma; DIRECTION: ltr; COLOR: rgb(0,0,0); FONT-=
SIZE: 10pt">
<p>&nbsp;</p>
<p>Although having the above concerns, I &nbsp;think we should: a) research=
 on certain dataset. It&nbsp;would be very useful to prove the machine lear=
ning could perform better or provide more flexibility and adaptibility;
</p>
</div>
</blockquote>
<div dir=3D"ltr" class=3D"gmail_quote"><br>
</div>
<div dir=3D"ltr" class=3D"gmail_quote">Not sure what you mean here. By &quo=
t;certain dataset&quot; you mean what?</div>
<div dir=3D"ltr" class=3D"gmail_quote">&nbsp;</div>
<div dir=3D"ltr" class=3D"gmail_quote">&lt;Sheng&gt;We did discuss a potent=
ial standardized dataset, didn't we?&lt;/Sheng&gt;</div>
<div dir=3D"ltr" class=3D"gmail_quote">&nbsp;</div>
<blockquote style=3D"BORDER-LEFT: rgb(204,204,204) 1px solid; MARGIN: 0px 0=
px 0px 0.8ex; PADDING-LEFT: 1ex" dir=3D"ltr" class=3D"gmail_quote">
<div style=3D"FONT-FAMILY: Tahoma; DIRECTION: ltr; COLOR: rgb(0,0,0); FONT-=
SIZE: 10pt">
<p>b) continue our discuss on various network use cases that could benefit =
from applying machine learning, in order to serve wider network communities=
.</p>
</div>
</blockquote>
<div dir=3D"ltr" class=3D"gmail_quote"><br>
</div>
<div dir=3D"ltr" class=3D"gmail_quote">Use cases are good, however machine =
learning isn't something we can just apply to a use case because we've defi=
ned the use case. There are really at least three parts of this that need t=
o be considered as part of a solution:
 (i). use case (as you mention), (ii). algorithm(s) that are going to provi=
de the regression/classification/ desired in (i), and (iii). data sets.&nbs=
p;</div>
<div dir=3D"ltr" class=3D"gmail_quote"><br>
</div>
<div dir=3D"ltr" class=3D"gmail_quote">(ii). requires understanding of the =
algorithms that can be applied to produce the inferences (regressions, ...)=
 &nbsp;we are interested in. What algorithms we can use depend on the data =
we can get (iii). &nbsp;</div>
<div dir=3D"ltr" class=3D"gmail_quote">&nbsp;</div>
<div dir=3D"ltr" class=3D"gmail_quote">&lt;Sheng&gt;What we have done up to=
 now is, in a specific use case, to design an algorithm or a combination of=
 algorithm on chosen dataset (this may be dynamic data from the network env=
ironment) to complete a specific task.&lt;/Sheng&gt;</div>
</div>
<div>
<div dir=3D"ltr" class=3D"gmail_quote">&nbsp;</div>
<div dir=3D"ltr" class=3D"gmail_quote">That said, one of the things we've l=
earned over the past decade or so is that learning itself is a very general=
 process. We see this in how powerful something as simple as SGD/backprop i=
s (see e.g., [2]). We can also use this
 to our advantage and re-purpose algorithms originally developed for other =
tasks. A simple example of this is LDA (see [3,4]). Here the original idea =
was to model topics in a corpus of documents, where the topics themselves (=
along with the per-document topic
 proportions) were the latent/hidden variables. One way to re-purpose LDA f=
or applications like ours (essentially classification/anomaly detection tas=
ks) is to view each to/from pair of IP addresses as a &quot;document&quot; =
(in the language of LDA) and then words are
 constructed from features of the to/from traffic (e.g., to ports, etc). Se=
e [5] for a decent example of how you might think about this kind of &quot;=
statistical clustering&quot; using the LDA Probabilistic Graphical Model ap=
proach. So there is hope, but there is a lot
 of work left to be done on both the theory and practice sides of this (whi=
ch as I mentioned earlier are, at least currently, essentially the same thi=
ng).</div>
<div dir=3D"ltr" class=3D"gmail_quote">&nbsp;</div>
<div dir=3D"ltr" class=3D"gmail_quote">&lt;Sheng&gt;Agree on that learning =
and learning algorithm is general. But what we did was only use it in speci=
fic way to solve specific tasks/use cases. As I agreed earlier, generlized =
learning on network could be an important
 direction and I am interesting in. Let's see how far we could go. On anoth=
er side, I have to emphasis that the general learning must be helpful on so=
lving specific tasks, in other words, specific tasks should be earlier to s=
olve from the result of general
 learning. Otherwise, the general learning is not mearningful.&lt;/Sheng&gt=
;</div>
<div dir=3D"ltr" class=3D"gmail_quote">&nbsp;</div>
<blockquote style=3D"BORDER-LEFT: rgb(204,204,204) 1px solid; MARGIN: 0px 0=
px 0px 0.8ex; PADDING-LEFT: 1ex" dir=3D"ltr" class=3D"gmail_quote">
<div style=3D"FONT-FAMILY: Tahoma; DIRECTION: ltr; COLOR: rgb(0,0,0); FONT-=
SIZE: 10pt">
<p>Another important point I would like to discuss is that you mentioned &q=
uot;our goal is to provide accurate, repeatable, and explainable results.&q=
uot; Ideally, we do want this. It is our traditional way for logic-based pr=
ogramming.</p>
</div>
</blockquote>
<div dir=3D"ltr" class=3D"gmail_quote">What is &quot;logic-based programmin=
g&quot;?&nbsp; Do you mean logic programming, e.g., Prolog? If so I underst=
and logic programming (I did a lot of work on logic programming back in the=
 late 80s/early 90s, see&nbsp;<a href=3D"http://dl.acm.org/citation.cfm?id=
=3D87991" target=3D"_blank">http://dl.acm.org/citation.cfm?id=3D87991</a>).
 But even so, I don't understand the connection, so can you be more explici=
t?</div>
<div dir=3D"ltr" class=3D"gmail_quote">&nbsp;</div>
<div dir=3D"ltr" class=3D"gmail_quote">&lt;Sheng&gt;Here, &quot;logic-based=
 programming&quot; I referred to all traditional programming that realize h=
uman pre-designed logic into computing programs. It is comparing to ML-base=
d programming.&lt;/Sheng&gt;</div>
<div dir=3D"ltr" class=3D"gmail_quote">&nbsp;</div>
<blockquote style=3D"BORDER-LEFT: rgb(204,204,204) 1px solid; MARGIN: 0px 0=
px 0px 0.8ex; PADDING-LEFT: 1ex" dir=3D"ltr" class=3D"gmail_quote">
<div style=3D"FONT-FAMILY: Tahoma; DIRECTION: ltr; COLOR: rgb(0,0,0); FONT-=
SIZE: 10pt">
<p>However, I am not sure that is a doable target&nbsp;with machine learnin=
g. Many of machine learning algorithms&nbsp;are based on statistics theory,=
 which may not be that much accurate.
</p>
</div>
</blockquote>
<div dir=3D"ltr" class=3D"gmail_quote">Sorry, I don't understand this. Thin=
gs like Bayesian inference (e.g., LDA) are inherently about quantifying unc=
ertainty. In any event can you be more specific? A learning algorithm can &=
quot;accurately&quot; approximate something like
 a posterior distribution, its just that the output is just that, a posteri=
or distribution. So there is a difference between &quot;accuracy&quot; (whi=
ch isn't really defined here) and the stochastic nature of the results we g=
et from ML.</div>
<div dir=3D"ltr" class=3D"gmail_quote">&nbsp;</div>
<div dir=3D"ltr" class=3D"gmail_quote">OTOH, if you mean accuracy of our es=
timated distributions then there are obviously ways this can go wrong. For =
example, even an algorithm as simple as k-means can go wrong since the opti=
mization objective (distortion function)
 is non-convex, so it is sensitive to initial conditions (in particular, wh=
ere you start; if you are unlucky you can get stuck in a local minimum &nbs=
p;due to the non-convex nature of the objective). See e.g., [8]. BTW, &nbsp=
;k-means was called out in one of the decks
 below but as I mentioned, not much discussion of the important topics we a=
re discussing here there. There is also approximate inference (for example,=
 MCMC or variational approaches), but these approaches also have provably&n=
bsp;quantitative error bounds (<a href=3D"http://www.1-4-5.net/~dmm/ml/ps.p=
df" target=3D"_blank">http://www.1-4-5.net/~dmm/ml/ps.pdf</a>
 talks a bit about how variational approaches, essential trading inference =
for optimization where the inference (integral) is intractable, find a tigh=
t lower-bound using Jensen's inequality.&nbsp;</div>
<div dir=3D"ltr" class=3D"gmail_quote">&nbsp;</div>
<div dir=3D"ltr" class=3D"gmail_quote">
<div dir=3D"ltr" class=3D"gmail_quote">&lt;Sheng&gt;The reason that statist=
ics-based ML algorithms can work is that they chose to focus on only high p=
robability and igrone corner cases. But when mearningful corner cases happe=
n or we want to target the corner cases, machine
 learning may not accurately work. Again, &quot;accuracy&quot; itself does =
not properly defined. So, we may talk about very different things. On other=
 side, I also agree that accurate is not important topic here. ML can reach=
 a high-level accurate, which are suffient
 for majority cases. But ML may never reach full accurate while traditional=
 program can 100% certain the result on giving circumstance.&lt;/Sheng&gt;<=
/div>
<div dir=3D"ltr" class=3D"gmail_quote">&nbsp;</div>
</div>
<blockquote style=3D"BORDER-LEFT: rgb(204,204,204) 1px solid; MARGIN: 0px 0=
px 0px 0.8ex; PADDING-LEFT: 1ex" dir=3D"ltr" class=3D"gmail_quote">
<div style=3D"FONT-FAMILY: Tahoma; DIRECTION: ltr; COLOR: rgb(0,0,0); FONT-=
SIZE: 10pt">
<p>Some of machine learning algorithms&nbsp;are black box or have very comp=
lexicated internal computing process. So, the resulf of these algorithms&nb=
sp;may be hard to be explained.
</p>
</div>
</blockquote>
<div dir=3D"ltr" class=3D"gmail_quote">This is definitely true to some exte=
nt today with DNNs, but that is changing rapidly. Results can be repeatable=
 but explaining exactly what units (artificial neurons) far from the input =
are doing can be difficult. But this
 too is an area of active research and development (see [6] from last year'=
s ICML, or the work I pointed to on GANs in my last note as progress in thi=
s area). Other approaches such probabilistic graphical models or signal pro=
cessing approaches are perhaps more
 easily explained, but I don't think this is the level of explanation that =
is going to be useful to people deploying and operating networks, so we'll =
have to find different ways of explaining what things like DNNs to in a way=
 that is consumable by operators,
 etc.</div>
<div dir=3D"ltr" class=3D"gmail_quote">&nbsp;</div>
<div dir=3D"ltr" class=3D"gmail_quote">&lt;Sheng&gt;Sharing the same opinio=
n here. Every we could really explain how DNNs works in a very complicated =
way. Our customers would not have enough patient to listen. They would stil=
l not trust ML. We, actually the whole AI/ML
 society, needs to find other way to explain and prove things like DNN are =
trustable. Only then, they can be use for more serious or autonomic&nbsp;ta=
sks.&lt;/Sheng&gt;</div>
<div dir=3D"ltr" class=3D"gmail_quote"><br>
</div>
<div dir=3D"ltr" class=3D"gmail_quote">All this by way of saying that we ar=
e gaining better understanding how DNNs work every day (literally) and as a=
 result &nbsp;I'm optimistic that we can build technology that will be accu=
rate (again, stochastic results), repeatable
 and explainable.</div>
<div dir=3D"ltr" class=3D"gmail_quote">&nbsp;</div>
<div dir=3D"ltr" class=3D"gmail_quote">&lt;Sheng&gt;I hope I can share you =
optimistic. Actually, I do believe we could do so one day. But I am not sur=
e how long it will take. &lt;/Sheng&gt;</div>
<div dir=3D"ltr" class=3D"gmail_quote"><br>
</div>
<div dir=3D"ltr" class=3D"gmail_quote">I'll just note here that if you have=
 a model with literally billions of parameters being trained by SGD its not=
 that surprising that humans have a hard time understanding what the comput=
ation is. As a concrete example, DNNs
 can solve the so-called &quot;selectivity-invariance&quot; problem [7], bu=
t we humans don't really &nbsp;know how to do, so this is another example.&=
nbsp;</div>
<div dir=3D"ltr" class=3D"gmail_quote">&nbsp;</div>
<blockquote style=3D"BORDER-LEFT: rgb(204,204,204) 1px solid; MARGIN: 0px 0=
px 0px 0.8ex; PADDING-LEFT: 1ex" dir=3D"ltr" class=3D"gmail_quote">
<div style=3D"FONT-FAMILY: Tahoma; DIRECTION: ltr; COLOR: rgb(0,0,0); FONT-=
SIZE: 10pt">
<p>Also the process of machine learning may be hard to be controlled or int=
ervened.
</p>
</div>
</blockquote>
<div dir=3D"ltr" class=3D"gmail_quote"><br>
</div>
<div dir=3D"ltr" class=3D"gmail_quote">Again, not sure what you mean here. =
Please give a bit more detail.</div>
<div dir=3D"ltr" class=3D"gmail_quote">&nbsp;</div>
<div dir=3D"ltr" class=3D"gmail_quote">&lt;Sheng&gt; Here, I actually refer=
 to DNN too. We don't know how to intervened DNN in its running time.</div>
<div dir=3D"ltr" class=3D"gmail_quote">&nbsp;</div>
<blockquote style=3D"BORDER-LEFT: rgb(204,204,204) 1px solid; MARGIN: 0px 0=
px 0px 0.8ex; PADDING-LEFT: 1ex" dir=3D"ltr" class=3D"gmail_quote">
<div style=3D"FONT-FAMILY: Tahoma; DIRECTION: ltr; COLOR: rgb(0,0,0); FONT-=
SIZE: 10pt">
<p>Repeatable, yes, we have to have this feature to be widely adopted. Howe=
ver, repeatable may have to come with some level of toleration for the inac=
curate.</p>
</div>
</blockquote>
<div dir=3D"ltr" class=3D"gmail_quote"><br>
</div>
<div dir=3D"ltr" class=3D"gmail_quote">All of this is inherently probabilis=
tic. A DNN, for example, can output the posterior distribution over some hi=
dden variables of interest accurately. LDA finds the latent &quot;topics&qu=
ot; and word distributions, both of which are
 multinomals so again we're essentially getting a posterior over the hidden=
 variables.&nbsp; Even k-means is &nbsp;just the EM (Expectation Maximizati=
on) algorithm applied to a particular Naive Bayes model. The same can be sa=
id of almost every learning algorithm you
 might mention. So &quot;inaccurate&quot; isn't the correct word here (unle=
ss you mean whatever you trained didn't predict whatever you wanted to pred=
ict, in which case something else is wrong). So the DNN gives you as stocha=
stic result, which may be different than what
 many network people are used to. But its not inaccurate, its just stochast=
ic.</div>
<div dir=3D"ltr" class=3D"gmail_quote">&nbsp;</div>
<div dir=3D"ltr" class=3D"gmail_quote">&lt;Sheng&gt;Inaccurate or stochasti=
c, they are not what many network people can accept. In their eyes, network=
 is definite, every parameter is definite. However, actually, if you could =
analyze every network parameter, the value
 may be definite, but actually they all have toleration for some level of i=
naccurate as long as their design purposes could meet. It is the fundementa=
l that we could use ML to solve network task.</div>
<div dir=3D"ltr" class=3D"gmail_quote">&nbsp;</div>
<div dir=3D"ltr" class=3D"gmail_quote">Thanks so much. Regards,</div>
<div dir=3D"ltr" class=3D"gmail_quote">&nbsp;</div>
<div dir=3D"ltr" class=3D"gmail_quote">Sheng &lt;/Sheng&gt;</div>
<div dir=3D"ltr" class=3D"gmail_quote">&nbsp;</div>
<div dir=3D"ltr" class=3D"gmail_quote"><br>
&nbsp;</div>
<div dir=3D"ltr" class=3D"gmail_quote">Thx,</div>
<div dir=3D"ltr" class=3D"gmail_quote"><br>
</div>
<div dir=3D"ltr" class=3D"gmail_quote">--dmm</div>
<div dir=3D"ltr" class=3D"gmail_quote"><br>
</div>
<div dir=3D"ltr" class=3D"gmail_quote"><br>
</div>
<div dir=3D"ltr" class=3D"gmail_quote"><br>
</div>
<div dir=3D"ltr" class=3D"gmail_quote">[0]&nbsp;<a href=3D"http://people.cs=
ail.mit.edu/mcollins/papers/CVPR_2008.pdf" target=3D"_blank">http://people.=
csail.mit.edu/mcollins/papers/CVPR_2008.pdf</a></div>
<div dir=3D"ltr" class=3D"gmail_quote">[1] &nbsp;<span style=3D"FONT-FAMILY=
: Tahoma; COLOR: rgb(0,0,0); FONT-SIZE: 10pt"><a href=3D"https://arxiv.org/=
abs/1206.5538" target=3D"_blank">https://arxiv.org/abs/1206.5538</a></span>=
</div>
<div dir=3D"ltr" class=3D"gmail_quote"><span style=3D"FONT-FAMILY: Tahoma; =
COLOR: rgb(0,0,0); FONT-SIZE: 10pt">[2]&nbsp;</span><font color=3D"#000000"=
 face=3D"Tahoma"><span style=3D"FONT-SIZE: 13px"><a href=3D"http://deliprao=
.com/archives/153" target=3D"_blank">http://deliprao.com/archives/153</a></=
span></font></div>
<div dir=3D"ltr" class=3D"gmail_quote"><font color=3D"#000000" face=3D"Taho=
ma"><span style=3D"FONT-SIZE: 13px">[3]&nbsp;<a href=3D"http://www.1-4-5.ne=
t/~dmm/ml/lda_intro.pdf" target=3D"_blank">http://www.1-4-5.net/~dmm/ml/lda=
_intro.pdf</a></span></font></div>
<div dir=3D"ltr" class=3D"gmail_quote"><font color=3D"#000000" face=3D"Taho=
ma"><span style=3D"FONT-SIZE: 13px">[4]&nbsp;<a href=3D"http://www.jmlr.org=
/papers/volume3/blei03a/blei03a.pdf" target=3D"_blank">http://www.jmlr.org/=
papers/volume3/blei03a/blei03a.pdf</a></span></font></div>
<div dir=3D"ltr" class=3D"gmail_quote"><font color=3D"#000000" face=3D"Taho=
ma"><span style=3D"FONT-SIZE: 13px">[5]&nbsp;<a href=3D"http://ieeexplore.i=
eee.org/stamp/stamp.jsp?arnumber=3D5949386" target=3D"_blank">http://ieeexp=
lore.ieee.org/stamp/stamp.jsp?arnumber=3D5949386</a></span></font></div>
<div dir=3D"ltr" class=3D"gmail_quote"><font color=3D"#000000" face=3D"Taho=
ma"><span style=3D"FONT-SIZE: 13px">[6]&nbsp;<a href=3D"http://yosinski.com=
/media/papers/Yosinski__2015__ICML_DL__Understanding_Neural_Networks_Throug=
h_Deep_Visualization__.pdf" target=3D"_blank">http://yosinski.com/media/pap=
ers/Yosinski__2015__ICML_DL__Understanding_Neural_Networks_Through_Deep_Vis=
ualization__.pdf</a></span></font></div>
<div dir=3D"ltr" class=3D"gmail_quote"><font color=3D"#000000" face=3D"Taho=
ma"><span style=3D"FONT-SIZE: 13px">[7]&nbsp;<a href=3D"https://cbmm.mit.ed=
u/sites/default/files/publications/Anselmi_Invariance_CBMM_memo_29.pdf" tar=
get=3D"_blank">https://cbmm.mit.edu/sites/default/files/publications/Anselm=
i_Invariance_CBMM_memo_29.pdf</a></span></font></div>
<div dir=3D"ltr" class=3D"gmail_quote"><font color=3D"#000000" face=3D"Taho=
ma"><span style=3D"FONT-SIZE: 13px">[8]&nbsp;<a href=3D"http://cs229.stanfo=
rd.edu/notes/cs229-notes7a.pdf" target=3D"_blank">http://cs229.stanford.edu=
/notes/cs229-notes7a.pdf</a></span></font></div>
<blockquote style=3D"BORDER-LEFT: rgb(204,204,204) 1px solid; MARGIN: 0px 0=
px 0px 0.8ex; PADDING-LEFT: 1ex" dir=3D"ltr" class=3D"gmail_quote">
<div>
<div style=3D"DIRECTION: ltr">
<p style=3D"FONT-FAMILY: Tahoma; COLOR: rgb(0,0,0); FONT-SIZE: 10pt">Best r=
egards,</p>
<p style=3D"FONT-FAMILY: Tahoma; COLOR: rgb(0,0,0); FONT-SIZE: 10pt">&nbsp;=
</p>
<p style=3D"FONT-FAMILY: Tahoma; COLOR: rgb(0,0,0); FONT-SIZE: 10pt">Sheng<=
/p>
<p style=3D"FONT-FAMILY: Tahoma; COLOR: rgb(0,0,0); FONT-SIZE: 10pt">&nbsp;=
</p>
<p style=3D"FONT-FAMILY: Tahoma; COLOR: rgb(0,0,0); FONT-SIZE: 10pt"></p>
<hr style=3D"FONT-FAMILY: Tahoma; COLOR: rgb(0,0,0); FONT-SIZE: 10pt">
<p style=3D"FONT-FAMILY: Tahoma; COLOR: rgb(0,0,0); FONT-SIZE: 10pt"></p>
<div>
<div style=3D"DIRECTION: ltr"><font color=3D"#000000" size=3D"2" face=3D"Ta=
homa"><b style=3D"FONT-FAMILY: 'Times New Roman'; COLOR: rgb(0,0,0); FONT-S=
IZE: 16px">From:</b><font face=3D"Times New Roman"><span style=3D"FONT-SIZE=
: 16px"> David Meyer [</span></font><a style=3D"FONT-FAMILY: 'Times New Rom=
an'; COLOR: rgb(0,0,0); FONT-SIZE: 16px" href=3D"mailto:dmm@1-4-5.net" targ=
et=3D"_blank">dmm@1-4-5.net</a><font face=3D"Times New Roman"><span style=
=3D"FONT-SIZE: 16px">]</span></font><br>
<b style=3D"FONT-FAMILY: 'Times New Roman'; COLOR: rgb(0,0,0); FONT-SIZE: 1=
6px">Sent:</b><font face=3D"Times New Roman"><span style=3D"FONT-SIZE: 16px=
"> 28 June 2016 23:30</span></font><a href=3D"http://deliprao.com/archives/=
153" target=3D"_blank">http://deliprao.com/archives/153</a><br>
<b style=3D"FONT-FAMILY: 'Times New Roman'; COLOR: rgb(0,0,0); FONT-SIZE: 1=
6px">To:</b><font face=3D"Times New Roman"><span style=3D"FONT-SIZE: 16px">=
 Sheng Jiang</span></font><br>
<b style=3D"FONT-FAMILY: 'Times New Roman'; COLOR: rgb(0,0,0); FONT-SIZE: 1=
6px">Cc:</b>
<a style=3D"FONT-FAMILY: 'Times New Roman'; COLOR: rgb(0,0,0); FONT-SIZE: 1=
6px" href=3D"mailto:nmlrg@irtf.org" target=3D"_blank">
nmlrg@irtf.org</a><br>
<b style=3D"FONT-FAMILY: 'Times New Roman'; COLOR: rgb(0,0,0); FONT-SIZE: 1=
6px">Subject:</b><font face=3D"Times New Roman"><span style=3D"FONT-SIZE: 1=
6px"> Re: [Nmlrg] links to each slide of presentations//RE: slides of NMLRG=
 #3, June 27th, Athens</span></font><br>
</font><br>
</div>
<div style=3D"FONT-FAMILY: 'Times New Roman'; COLOR: rgb(0,0,0); FONT-SIZE:=
 16px">
<div class=3D"h5">
<div></div>
<div>
<div dir=3D"ltr">Sheng,
<div><br>
</div>
<div>Thanks for the pointers. One thing I notice is that we don't have much=
 on what the characteristics of network data might be, and as such, what ki=
nds of existing learning algorithms might be suitable and where additional =
development might be required. For
 example, is the collected data IID? Is it time series? Is the underlying d=
istribution stationary? If not, how does that constrain algorithms we might=
 use or develop? For example, how does a given algorithm deal with concept =
drift/internal covariate shift (in
 the case of DNNs; see e.g., batch normalization [0]). There are many other=
 such questions, such as is the data categorical (e.g., ports, IP addresses=
) or is it&nbsp;continuous/discrete (e.g., counters). And if the data in qu=
estion is categorical, what is the&nbsp;cardinality
 of the categories (this will inform how such data can be encoded); in the =
case of IP addresses we can't really one-hot encode addresses because their=
 cardinality is too large (2^32 or 2^128); this has implications for how we=
 build classifiers (in particular,
 for softmax layers in DNNs of various kinds).</div>
<div><br>
</div>
<div>Related to the above is the question of features? What are good featur=
es for networking? Where do they come from? Are they domain specific? Can w=
e learn features the way a DNNs does in the network space? Can we use autoe=
ncoders to discover such features?
 Or can we use GANs to train DNNs for network classification tasks in an un=
supervised manner? Are there other, non-ad-hoc (well founded) methods we ca=
n use, or is every use case a one-off (one would hope not).</div>
<div><br>
</div>
<div>We can carry the same kinds of analyzes to the algorithms applied. For=
 example, while something k-means is an effective way to get a feeling for =
how&nbsp;continuous/discrete hangs together, if our data is categorical sta=
tistical clustering approaches such as
 LDA might provide a more well-founded approach (of course, as with most Ba=
yesian techniques is the question of approximate inference arises since in =
most interesting cases the integral that we need to solve, namely the margi=
nal probability data isn't tractable
 so we need to resort to MCMC or more likely&nbsp;variational inference). A=
nd what about the use of SGD/batch normalization etc with DNNs, and perhaps=
 more importantly, can we use network data to train DNN policy networks for=
 reinforcement learning like we saw in
 deep Q-learning and AlphaGo?</div>
<div><br>
</div>
<div>These comments are all by way of saying that we don't have a solid the=
oretical understanding (yet) of how techniques that have been so successful=
 in other domains (e.g., DNNs for perceptual tasks) generalize to networkin=
g use cases. We will need this understanding
 if our goal is to provide accurate, repeatable, and explainable results.&n=
bsp;</div>
<div><br>
</div>
<div>In order to accomplish all of this we need, as I have been saying , no=
t only a good understanding of how these algorithms work but also standardi=
zed data sets and associated benchmarks so we can tell if we are making pro=
gress (or even if our techniques
 work). Analogies here include MNIST and ImageNet and their associated benc=
hmarks, among others. As mentioned &nbsp;standardized data sets are key to =
making progress in the ML for networking space (otherwise how do you know y=
our technique works and/or improves on
 another techniques?). One might assume that these data sets would need to =
be labeled (as supervised learning is where most of the progress is being m=
ade these days), but not necessarily; Generative Adversarial Networks (GANs=
) have emerged as a new way to train
 DNNs in an unsupervised manner (this is moving very rapidly; see e.g.,&nbs=
p;<a href=3D"https://openai.com/blog/generative-models/" target=3D"_blank">=
https://openai.com/blog/generative-models/</a>).</div>
<div><br>
</div>
<div>The summary here is that the &quot;distance&quot; between theory and p=
ractice in ML is effectively zero right now due to the incredible rate of p=
rogress in the field; this means we need &nbsp;to understand both sides of =
the theory/practice coin in order to be effective.
 None of the slide decks provide much background on what the proposed algor=
ithms are, how they work, or why they should be expected to work on network=
 data.</div>
<div><br>
</div>
<div>Finally, if you are interested in LDA or other algorithms there are a =
few short explanatory pieces I have written for my team on
<a href=3D"http://www.1-4-5.net/~dmm/ml" target=3D"_blank">http://www.1-4-5=
.net/~dmm/ml</a> (works in progress).</div>
<div><br>
</div>
<div>Thanks,</div>
<div><br>
Dave</div>
<div><br>
</div>
<div>[0]&nbsp;<a href=3D"https://arxiv.org/pdf/1502.03167.pdf" target=3D"_b=
lank">https://arxiv.org/pdf/1502.03167.pdf</a></div>
</div>
<div class=3D"gmail_extra"><br>
<div class=3D"gmail_quote">On Tue, Jun 28, 2016 at 7:03 AM, Sheng Jiang <sp=
an dir=3D"ltr">
&lt;<a href=3D"mailto:jiangsheng@huawei.com" target=3D"_blank">jiangsheng@h=
uawei.com</a>&gt;</span> wrote:<br>
<blockquote style=3D"BORDER-LEFT: rgb(204,204,204) 1px solid; MARGIN: 0px 0=
px 0px 0.8ex; PADDING-LEFT: 1ex" class=3D"gmail_quote">
Oops... The proceeding page for interim meeting seems not be as intelligent=
 as the proceeding pages of IETF meetings. Our proceeding page does not aut=
onomically show the slides. I have sent an email to IETF secretary to ask t=
hem to fix it. Meanwhile, in this
 email, here is the links for each presentations:<br>
<br>
Slide Filename Edit Replace Delete<br>
Chair Slides<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-0.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-0.pdf</a><br>
<br>
Introduction to Network Machine Learning &amp; NMLRG<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-1.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-1.pdf</a><br>
<br>
Data Collection and Analysis At High Security Lab<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-2.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-2.pdf</a><br>
<br>
Use Cases of Applying Machine Learning Mechanism with Network Traffic<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-3.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-3.pdf</a><br>
<br>
Mobile network state characterization and prediction<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-4.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-4.pdf</a><br>
<br>
Learning how to route<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-5.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-5.pdf</a><br>
<br>
Regards,<br>
<br>
Sheng<br>
________________________________________<br>
From: nmlrg [<a href=3D"mailto:nmlrg-bounces@irtf.org" target=3D"_blank">nm=
lrg-bounces@irtf.org</a>] on behalf of Sheng Jiang [<a href=3D"mailto:jiang=
sheng@huawei.com" target=3D"_blank">jiangsheng@huawei.com</a>]<br>
Sent: 28 June 2016 21:03<br>
To: <a href=3D"mailto:nmlrg@irtf.org" target=3D"_blank">nmlrg@irtf.org</a><=
br>
Subject: [Nmlrg] slides of NMLRG #3, June 27th, Athens<br>
<br>
Hi, nmlrg,<br>
<br>
All slides that have been presented in our NMLRG #3 meeting, June 27th, 201=
6, Athens, Greece, with EUCNC2016, have been uploaded. They can be accessed=
 through below link<br>
<br>
<a href=3D"https://www.ietf.org/proceedings/interim/2016/06/27/nmlrg/procee=
dings.html" rel=3D"noreferrer" target=3D"_blank">https://www.ietf.org/proce=
edings/interim/2016/06/27/nmlrg/proceedings.html</a><br>
<br>
Best regards,<br>
<br>
Sheng<br>
_______________________________________________<br>
nmlrg mailing list<br>
<a href=3D"mailto:nmlrg@irtf.org" target=3D"_blank">nmlrg@irtf.org</a><br>
<a href=3D"https://www.irtf.org/mailman/listinfo/nmlrg" rel=3D"noreferrer" =
target=3D"_blank">https://www.irtf.org/mailman/listinfo/nmlrg</a><br>
_______________________________________________<br>
nmlrg mailing list<br>
<a href=3D"mailto:nmlrg@irtf.org" target=3D"_blank">nmlrg@irtf.org</a><br>
<a href=3D"https://www.irtf.org/mailman/listinfo/nmlrg" rel=3D"noreferrer" =
target=3D"_blank">https://www.irtf.org/mailman/listinfo/nmlrg</a><br>
</blockquote>
</div>
<br>
</div>
</div>
</div>
</div>
</div>
</div>
</div>
</blockquote>
<div dir=3D"ltr" class=3D"gmail_extra"><br>
</div>
</div>
</div>
</div>
</body>
</html>

--_000_5D36713D8A4E7348A7E10DF7437A4B927CA89FB4NKGEML515MBXchi_--


From nobody Thu Jun 30 13:52:51 2016
Return-Path: <dmm@1-4-5.net>
X-Original-To: nmlrg@ietfa.amsl.com
Delivered-To: nmlrg@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 17A6512B01C for <nmlrg@ietfa.amsl.com>; Thu, 30 Jun 2016 13:52:50 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.599
X-Spam-Level: 
X-Spam-Status: No, score=-2.599 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, HTML_MESSAGE=0.001, RCVD_IN_DNSWL_LOW=-0.7] autolearn=ham autolearn_force=no
Authentication-Results: ietfa.amsl.com (amavisd-new); dkim=pass (2048-bit key) header.d=1-4-5-net.20150623.gappssmtp.com
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id N0jYd_1nUiKs for <nmlrg@ietfa.amsl.com>; Thu, 30 Jun 2016 13:52:42 -0700 (PDT)
Received: from mail-io0-x234.google.com (mail-io0-x234.google.com [IPv6:2607:f8b0:4001:c06::234]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by ietfa.amsl.com (Postfix) with ESMTPS id 2E369128874 for <nmlrg@irtf.org>; Thu, 30 Jun 2016 13:52:42 -0700 (PDT)
Received: by mail-io0-x234.google.com with SMTP id s63so83250437ioi.3 for <nmlrg@irtf.org>; Thu, 30 Jun 2016 13:52:42 -0700 (PDT)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1-4-5-net.20150623.gappssmtp.com; s=20150623; h=mime-version:in-reply-to:references:from:date:message-id:subject:to :cc; bh=Sr8efkqJX7UaxftojYeHtLk28RVGgfyFNb8Nzni1Pv0=; b=yxd7jZ07Idr5S/LCu+PdA4x/kT6FVdALqKExrKvjM3LE5NqXAtVOROZfYASRVqvwgx sQ9LiFMjpdVJzJ8rXTakGI50wNSp9J2oFO9d6eM2sgCnVPeltrA2pMyIUQYUPhxRMTbV xLw7gHcP8JRhGpA3Ss0Uo35+VumA1s0dHvvcdXdSKSlfVRGmVBmcx2rDWsvt5/Z4xn5V D1dhSjzQ09zitV9ayQxZmBDQy2d2ya6jzpSbUHsxkwtWDYD5rq9W70CHYCG/Jc/0xBmn mfFGJ4CKAKXmWqwCcrqVbjdD82w1YiMc86h968cpgJneOXMkzCmTeW+YGP+1nGZz5nna fAHw==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:from:date :message-id:subject:to:cc; bh=Sr8efkqJX7UaxftojYeHtLk28RVGgfyFNb8Nzni1Pv0=; b=FyFZKb5/xRlwRESDqIsWMLFEmvze6qsmoEXmclCIkI/pDoEbmLHNALgpQJSqkUscOY x9mvXRBxzmgSXaJ+a/BNmauWbsH3phZTPAsWL2ZhlOw7fCOjO7lRpfTXqazTAhNgsMLF KMHwZQD8EFUBdYAasgrMxM62F5lWabvwktC5J0h1TQgpletjuQY7QzNmd0875Se3w5Vm onA0SVnkkBRE/gpH4Ry3gEitANTRrmzsB+pBDqdoaxsGkQLSEfY6SdEKZbyxdVqmXvuH 8jDRjRefHATk9rl2KJk/tzAzhWEgrAS+my6jd86O9S4UhnkmttRo/8msgiiSLZWmlK2G MO4w==
X-Gm-Message-State: ALyK8tJAnAaKAe1jaCefrS7Da/tOhkecvC6dtErVEtDV1Z7lM9z7QBoIS7MXtA+8nLKL5tvSP61aWjKcbhZ0Og==
X-Received: by 10.107.29.142 with SMTP id d136mr1686575iod.50.1467319961357; Thu, 30 Jun 2016 13:52:41 -0700 (PDT)
MIME-Version: 1.0
Received: by 10.64.148.133 with HTTP; Thu, 30 Jun 2016 13:52:39 -0700 (PDT)
X-Originating-IP: [67.189.0.51]
In-Reply-To: <5D36713D8A4E7348A7E10DF7437A4B927CA89FB4@NKGEML515-MBX.china.huawei.com>
References: <5D36713D8A4E7348A7E10DF7437A4B927CA893E9@NKGEML515-MBX.china.huawei.com> <CAHiKxWhfdukjRVnSKnbLwMwamiqZYoCD350BoNqO4njqkQoU1g@mail.gmail.com> <5D36713D8A4E7348A7E10DF7437A4B927CA897D4@NKGEML515-MBX.china.huawei.com> <CAHiKxWhMPjKY24zucg-UND0ARWHbjnOyYv=Tia-yk-Vq6Ta_SA@mail.gmail.com> <5D36713D8A4E7348A7E10DF7437A4B927CA89FB4@NKGEML515-MBX.china.huawei.com>
From: David Meyer <dmm@1-4-5.net>
Date: Thu, 30 Jun 2016 13:52:39 -0700
Message-ID: <CAHiKxWh5D0JiDLTKV3ePT4XNhk9ZhmDu66Z=5hxOpoo5pxkorA@mail.gmail.com>
To: Sheng Jiang <jiangsheng@huawei.com>
Content-Type: multipart/alternative; boundary=001a1140958411a5e30536850d34
Archived-At: <https://mailarchive.ietf.org/arch/msg/nmlrg/hJgKBvfVUPNoZLB79culrGxlALk>
Cc: "nmlrg@irtf.org" <nmlrg@irtf.org>
Subject: Re: [Nmlrg] links to each slide of presentations//RE: slides of NMLRG #3, June 27th, Athens
X-BeenThere: nmlrg@irtf.org
X-Mailman-Version: 2.1.17
Precedence: list
List-Id: Network Machine Learning Research Group <nmlrg.irtf.org>
List-Unsubscribe: <https://www.irtf.org/mailman/options/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=unsubscribe>
List-Archive: <https://mailarchive.ietf.org/arch/browse/nmlrg/>
List-Post: <mailto:nmlrg@irtf.org>
List-Help: <mailto:nmlrg-request@irtf.org?subject=help>
List-Subscribe: <https://www.irtf.org/mailman/listinfo/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=subscribe>
X-List-Received-Date: Thu, 30 Jun 2016 20:52:50 -0000

--001a1140958411a5e30536850d34
Content-Type: text/plain; charset=UTF-8

Hey Sheng, see in-line. Thx, --dmm

On Thu, Jun 30, 2016 at 1:16 PM, Sheng Jiang <jiangsheng@huawei.com> wrote:

> Hi, David,
>
>
>
> Please see my replies with <Sheng/> in the lines. Sorry for being slow
> response due to my traveling.
>
>
>
>> However, upto now, I have serveral concerns. Firstly, before we could
>> discuss such dataset, we need to narrow our target scenarios to a few
>> very specific use cases, giving that there are so many various scenarios in
>> the network area and they are so different from each other and require
>> different dataset.
>>
>
> I'm hoping that not every network is not a one off. Rather, what we seek
> is to learn what are the fundamental properties of networks that can be the
> basis of reasoning about networks do. For example, we should be able to
> learn something from work done on transfer learning [0] and representation
> learning [1]. In the image recognition space there are even startups that
> provide pretrained AlexNet, GooLeNet, etc trained on ImageNet (and others)
> that you can further train/customize for your particular application. In
> any event, if every network requires something totally different then we
> haven't yet understood what is fundamental to the structure of our networks.
>
> <Sheng>If I understand you correctly, you meant we should do some generic
> learning in order to better understand the network fundamentally, and the
> result should be helpful for solving these specific tasks in the networks.
> It is different to what we have done - working on applying ML to solve
> specific network tasks. I would agree this proposed research direction are
> important and  I am very interesting in such direction although  it looks
> ambitions and no guarantee for success so far. </Sheng>
>

No, not really. What I meant is if every network is a one-off then by
definition we haven't learned to generalize. Now, generalization might not
be possible, but I don't see any reason why it wouldn't be. It certainly is
for perceptual tasks, so there's one data point (for example, if you want
to do image recognition you use a Conv Net because there is both theory and
practice behind the framework).


>
>> Secondly, as we did talk with operators, the real data is the important
>> properties for them. So, it would be almost impossible for them to share
>> them. It leaves two possibilities on how to get dataset: a) generating
>> dataset using simulation;
>>
>
> This likely won't work. For example, if you do this then try to train a
> DNN on that data set the NN will learn how the simulator works. This is
> unlikely to be the outcome that we'd like (we already know how the
> simulator works).
>
>  b) try to get some partial dataset from operators that have removed
> their sensenble data. Both way are problemstic, in the risk that a lot of
> efforts may waste in wrong areas.
>
> Agree that this is problematic.
>
> <Sheng>Here, we seems reach the agreement on that we do NOT have a good
> way to obtain the useful dataset. It is not good at all. Does that mean we
> have to wait for operators to work on themselves alone? Any suggestion on
> how we could get quality real dataset?</Sheng>
>

This is a huge challenge, and one we should spend time talking about.
Without something like ImageNet (or MNIST), we want to able to measure
progress (or even if our techniques work).


Thirdly, the availability of data may very different in different network
> environments. It depends on the measurement/congnitive functions that have
> been deployed or implemented.
>
> What is a "cognitive function"?
>
> <Sheng>Here, I use the "cognitive function" to represent that device or
> network function that collects the existing status or configuration data,
> against measurement.</Sheng>
>


So is a network element, say a router, a cognitive function?


> In any event, you have outlined just a few of the problems with creating
> "standardized" data sets for our use. I have quoted standardized because
> I'm beginning to think its the wrong word (invokes "standards" e.g., IETF,
> <X>..., and that is not what is meant here).  Not sure what the right term
> is but  "standardized" seems to be the wrong word here.
>
>
>>
>> Although having the above concerns, I  think we should: a) research on
>> certain dataset. It would be very useful to prove the machine learning
>> could perform better or provide more flexibility and adaptibility;
>>
>
> Not sure what you mean here. By "certain dataset" you mean what?
>
> <Sheng>We did discuss a potential standardized dataset, didn't we?</Sheng>
>

Sorry,  I didn't understand the language.


>
>
>> b) continue our discuss on various network use cases that could benefit
>> from applying machine learning, in order to serve wider network communities.
>>
>
> Use cases are good, however machine learning isn't something we can just
> apply to a use case because we've defined the use case. There are really at
> least three parts of this that need to be considered as part of a solution:
> (i). use case (as you mention), (ii). algorithm(s) that are going to
> provide the regression/classification/ desired in (i), and (iii). data
> sets.
>
> (ii). requires understanding of the algorithms that can be applied to
> produce the inferences (regressions, ...)  we are interested in. What
> algorithms we can use depend on the data we can get (iii).
>
> <Sheng>What we have done up to now is, in a specific use case, to design
> an algorithm or a combination of algorithm on chosen dataset (this may be
> dynamic data from the network environment) to complete a specific
> task.</Sheng>
>
> That said, one of the things we've learned over the past decade or so is
> that learning itself is a very general process. We see this in how powerful
> something as simple as SGD/backprop is (see e.g., [2]). We can also use
> this to our advantage and re-purpose algorithms originally developed for
> other tasks. A simple example of this is LDA (see [3,4]). Here the original
> idea was to model topics in a corpus of documents, where the topics
> themselves (along with the per-document topic proportions) were the
> latent/hidden variables. One way to re-purpose LDA for applications like
> ours (essentially classification/anomaly detection tasks) is to view each
> to/from pair of IP addresses as a "document" (in the language of LDA) and
> then words are constructed from features of the to/from traffic (e.g., to
> ports, etc). See [5] for a decent example of how you might think about this
> kind of "statistical clustering" using the LDA Probabilistic Graphical
> Model approach. So there is hope, but there is a lot of work left to be
> done on both the theory and practice sides of this (which as I mentioned
> earlier are, at least currently, essentially the same thing).
>
> <Sheng>Agree on that learning and learning algorithm is general. But what
> we did was only use it in specific way to solve specific tasks/use cases.
>

Understand, but I'm questioning whether this is a legitimate topic for a
research group. Building one-off solutions is obviously possible (look at
all of the startups, products etc in this space), but none of these
approachs give us insight into how to learn from network data, what its
properties are, or how to generalize (really the holy grail).

As I agreed earlier, generlized learning on network could be an important
> direction and I am interesting in. Let's see how far we could go. On
> another side, I have to emphasis that the general learning must be helpful
> on solving specific tasks, in other words, specific tasks should be earlier
> to solve from the result of general learning. Otherwise, the general
> learning is not mearningful.</Sheng>
>
>
>> Another important point I would like to discuss is that you mentioned
>> "our goal is to provide accurate, repeatable, and explainable results."
>> Ideally, we do want this. It is our traditional way for logic-based
>> programming.
>>
> What is "logic-based programming"?  Do you mean logic programming, e.g.,
> Prolog? If so I understand logic programming (I did a lot of work on logic
> programming back in the late 80s/early 90s, see
> http://dl.acm.org/citation.cfm?id=87991). But even so, I don't understand
> the connection, so can you be more explicit?
>
> <Sheng>Here, "logic-based programming" I referred to all traditional
> programming that realize human pre-designed logic into computing programs.
> It is comparing to ML-based programming.</Sheng>
>

None of those terms are well defined, but I get your sense.

>
>
>> However, I am not sure that is a doable target with machine learning.
>> Many of machine learning algorithms are based on statistics theory, which
>> may not be that much accurate.
>>
> Sorry, I don't understand this. Things like Bayesian inference (e.g., LDA)
> are inherently about quantifying uncertainty. In any event can you be more
> specific? A learning algorithm can "accurately" approximate something like
> a posterior distribution, its just that the output is just that, a
> posterior distribution. So there is a difference between "accuracy" (which
> isn't really defined here) and the stochastic nature of the results we get
> from ML.
>
> OTOH, if you mean accuracy of our estimated distributions then there are
> obviously ways this can go wrong. For example, even an algorithm as simple
> as k-means can go wrong since the optimization objective (distortion
> function) is non-convex, so it is sensitive to initial conditions (in
> particular, where you start; if you are unlucky you can get stuck in a
> local minimum  due to the non-convex nature of the objective). See e.g.,
> [8]. BTW,  k-means was called out in one of the decks below but as I
> mentioned, not much discussion of the important topics we are discussing
> here there. There is also approximate inference (for example, MCMC or
> variational approaches), but these approaches also have
> provably quantitative error bounds (http://www.1-4-5.net/~dmm/ml/ps.pdf
> talks a bit about how variational approaches, essential trading inference
> for optimization where the inference (integral) is intractable, find a
> tight lower-bound using Jensen's inequality.
>
> <Sheng>The reason that statistics-based ML algorithms can work is that
> they chose to focus on only high probability and igrone corner cases.
>

I don't believe this is correct, but perhaps I don't understand your point.
Can you give a concrete example of (a). what you mean by statistics-based
ML algorithm and (b). one that ignores corner cases and how/why?


> But when mearningful corner cases happen or we want to target the corner
> cases, machine learning may not accurately work. Again, "accuracy" itself
> does not properly defined. So, we may talk about very different things. On
> other side, I also agree that accurate is not important topic here. ML can
> reach a high-level accurate, which are suffient for majority cases. But ML
> may never reach full accurate while traditional program can 100% certain
> the result on giving circumstance.</Sheng>
>
>
>> Some of machine learning algorithms are black box or have very
>> complexicated internal computing process. So, the resulf of these
>> algorithms may be hard to be explained.
>>
> This is definitely true to some extent today with DNNs, but that is
> changing rapidly. Results can be repeatable but explaining exactly what
> units (artificial neurons) far from the input are doing can be difficult.
> But this too is an area of active research and development (see [6] from
> last year's ICML, or the work I pointed to on GANs in my last note as
> progress in this area). Other approaches such probabilistic graphical
> models or signal processing approaches are perhaps more easily explained,
> but I don't think this is the level of explanation that is going to be
> useful to people deploying and operating networks, so we'll have to find
> different ways of explaining what things like DNNs to in a way that is
> consumable by operators, etc.
>
> <Sheng>Sharing the same opinion here. Every we could really explain how
> DNNs works in a very complicated way. Our customers would not have enough
> patient to listen. They would still not trust ML. We, actually the whole
> AI/ML society, needs to find other way to explain and prove things like DNN
> are trustable. Only then, they can be use for more serious or
> autonomic tasks.</Sheng>
>

I'm optimistic that we will be able to understand what the higher layers in
DNNs are doing; much progress has already been and is being made. That
said, this problem is really more of an issue for consumers of ML
technology, and not so much for those building it (although it would be
useful to better understand the computation a DNN is making; as I mentioned
much progress has been made here). As a loose analogy, how many consumers
of the Linux Kernel really understand what it is doing in any detailed way?
Of course explainability is important and I don't mean to diminish that
importance in any way.

BTW, people are going to deploy ML based systems not because they are
interesting or cool (which they are), but rather because their competitors
are/will. We already see this in the hyper-scales, automotive industry,
process automation industry, and even our industry (check out Chris'
description of ML as "automating your automation"
https://www.youtube.com/watch?v=PZAAOVbiK9A).


> All this by way of saying that we are gaining better understanding how
> DNNs work every day (literally) and as a result  I'm optimistic that we can
> build technology that will be accurate (again, stochastic results),
> repeatable and explainable.
>
> <Sheng>I hope I can share you optimistic. Actually, I do believe we could
> do so one day. But I am not sure how long it will take. </Sheng>
>
> I'll just note here that if you have a model with literally billions of
> parameters being trained by SGD its not that surprising that humans have a
> hard time understanding what the computation is. As a concrete example,
> DNNs can solve the so-called "selectivity-invariance" problem [7], but we
> humans don't really  know how to do, so this is another example.
>
>
>> Also the process of machine learning may be hard to be controlled or
>> intervened.
>>
>
> Again, not sure what you mean here. Please give a bit more detail.
>
> <Sheng> Here, I actually refer to DNN too. We don't know how to intervened
> DNN in its running time.
>
>

Do you mean that once you've trained a DNN you can only use it for
classification (or whatever) and can't further train it, or do you mean
that we can't learn from additional data, or something else?


> Repeatable, yes, we have to have this feature to be widely adopted.
>> However, repeatable may have to come with some level of toleration for the
>> inaccurate.
>>
>
> All of this is inherently probabilistic. A DNN, for example, can output
> the posterior distribution over some hidden variables of interest
> accurately. LDA finds the latent "topics" and word distributions, both of
> which are multinomals so again we're essentially getting a posterior over
> the hidden variables.  Even k-means is  just the EM (Expectation
> Maximization) algorithm applied to a particular Naive Bayes model. The same
> can be said of almost every learning algorithm you might mention. So
> "inaccurate" isn't the correct word here (unless you mean whatever you
> trained didn't predict whatever you wanted to predict, in which case
> something else is wrong). So the DNN gives you as stochastic result, which
> may be different than what many network people are used to. But its not
> inaccurate, its just stochastic.
>
> <Sheng>Inaccurate or stochastic, they are not what many network people can
> accept. In their eyes, network is definite, every parameter is definite.
> However, actually, if you could analyze every network parameter, the value
> may be definite, but actually they all have toleration for some level of
> inaccurate as long as their design purposes could meet. It is the
> fundemental that we could use ML to solve network task.
>


I didn't understand what you mean here.

Thanks,

--dmm


>
> Thanks so much. Regards,
>
> Sheng </Sheng>
>
>
>
> Thx,
>
> --dmm
>
>
>
> [0] http://people.csail.mit.edu/mcollins/papers/CVPR_2008.pdf
> [1]  https://arxiv.org/abs/1206.5538
> [2] http://deliprao.com/archives/153
> [3] http://www.1-4-5.net/~dmm/ml/lda_intro.pdf
> [4] http://www.jmlr.org/papers/volume3/blei03a/blei03a.pdf
> [5] http://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=5949386
> [6]
> http://yosinski.com/media/papers/Yosinski__2015__ICML_DL__Understanding_Neural_Networks_Through_Deep_Visualization__.pdf
> [7]
> https://cbmm.mit.edu/sites/default/files/publications/Anselmi_Invariance_CBMM_memo_29.pdf
> [8] http://cs229.stanford.edu/notes/cs229-notes7a.pdf
>
>> Best regards,
>>
>>
>>
>> Sheng
>>
>>
>>
>> ------------------------------
>>
>> *From:* David Meyer [dmm@1-4-5.net]
>> *Sent:* 28 June 2016 23:30http://deliprao.com/archives/153
>> *To:* Sheng Jiang
>> *Cc:* nmlrg@irtf.org
>> *Subject:* Re: [Nmlrg] links to each slide of presentations//RE: slides
>> of NMLRG #3, June 27th, Athens
>>
>> Sheng,
>>
>> Thanks for the pointers. One thing I notice is that we don't have much on
>> what the characteristics of network data might be, and as such, what kinds
>> of existing learning algorithms might be suitable and where additional
>> development might be required. For example, is the collected data IID? Is
>> it time series? Is the underlying distribution stationary? If not, how does
>> that constrain algorithms we might use or develop? For example, how does a
>> given algorithm deal with concept drift/internal covariate shift (in the
>> case of DNNs; see e.g., batch normalization [0]). There are many other such
>> questions, such as is the data categorical (e.g., ports, IP addresses) or
>> is it continuous/discrete (e.g., counters). And if the data in question is
>> categorical, what is the cardinality of the categories (this will inform
>> how such data can be encoded); in the case of IP addresses we can't really
>> one-hot encode addresses because their cardinality is too large (2^32 or
>> 2^128); this has implications for how we build classifiers (in particular,
>> for softmax layers in DNNs of various kinds).
>>
>> Related to the above is the question of features? What are good features
>> for networking? Where do they come from? Are they domain specific? Can we
>> learn features the way a DNNs does in the network space? Can we use
>> autoencoders to discover such features? Or can we use GANs to train DNNs
>> for network classification tasks in an unsupervised manner? Are there
>> other, non-ad-hoc (well founded) methods we can use, or is every use case a
>> one-off (one would hope not).
>>
>> We can carry the same kinds of analyzes to the algorithms applied. For
>> example, while something k-means is an effective way to get a feeling for
>> how continuous/discrete hangs together, if our data is categorical
>> statistical clustering approaches such as LDA might provide a more
>> well-founded approach (of course, as with most Bayesian techniques is the
>> question of approximate inference arises since in most interesting cases
>> the integral that we need to solve, namely the marginal probability data
>> isn't tractable so we need to resort to MCMC or more likely variational
>> inference). And what about the use of SGD/batch normalization etc with
>> DNNs, and perhaps more importantly, can we use network data to train DNN
>> policy networks for reinforcement learning like we saw in deep Q-learning
>> and AlphaGo?
>>
>> These comments are all by way of saying that we don't have a solid
>> theoretical understanding (yet) of how techniques that have been so
>> successful in other domains (e.g., DNNs for perceptual tasks) generalize to
>> networking use cases. We will need this understanding if our goal is to
>> provide accurate, repeatable, and explainable results.
>>
>> In order to accomplish all of this we need, as I have been saying , not
>> only a good understanding of how these algorithms work but also
>> standardized data sets and associated benchmarks so we can tell if we are
>> making progress (or even if our techniques work). Analogies here include
>> MNIST and ImageNet and their associated benchmarks, among others. As
>> mentioned  standardized data sets are key to making progress in the ML for
>> networking space (otherwise how do you know your technique works and/or
>> improves on another techniques?). One might assume that these data sets
>> would need to be labeled (as supervised learning is where most of the
>> progress is being made these days), but not necessarily; Generative
>> Adversarial Networks (GANs) have emerged as a new way to train DNNs in an
>> unsupervised manner (this is moving very rapidly; see e.g.,
>> https://openai.com/blog/generative-models/).
>>
>> The summary here is that the "distance" between theory and practice in ML
>> is effectively zero right now due to the incredible rate of progress in the
>> field; this means we need  to understand both sides of the theory/practice
>> coin in order to be effective. None of the slide decks provide much
>> background on what the proposed algorithms are, how they work, or why they
>> should be expected to work on network data.
>>
>> Finally, if you are interested in LDA or other algorithms there are a few
>> short explanatory pieces I have written for my team on
>> http://www.1-4-5.net/~dmm/ml (works in progress).
>>
>> Thanks,
>>
>> Dave
>>
>> [0] https://arxiv.org/pdf/1502.03167.pdf
>>
>> On Tue, Jun 28, 2016 at 7:03 AM, Sheng Jiang <jiangsheng@huawei.com>
>> wrote:
>>
>>> Oops... The proceeding page for interim meeting seems not be as
>>> intelligent as the proceeding pages of IETF meetings. Our proceeding page
>>> does not autonomically show the slides. I have sent an email to IETF
>>> secretary to ask them to fix it. Meanwhile, in this email, here is the
>>> links for each presentations:
>>>
>>> Slide Filename Edit Replace Delete
>>> Chair Slides
>>>
>>> https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-0.pdf
>>>
>>> Introduction to Network Machine Learning & NMLRG
>>>
>>> https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-1.pdf
>>>
>>> Data Collection and Analysis At High Security Lab
>>>
>>> https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-2.pdf
>>>
>>> Use Cases of Applying Machine Learning Mechanism with Network Traffic
>>>
>>> https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-3.pdf
>>>
>>> Mobile network state characterization and prediction
>>>
>>> https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-4.pdf
>>>
>>> Learning how to route
>>>
>>> https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-5.pdf
>>>
>>> Regards,
>>>
>>> Sheng
>>> ________________________________________
>>> From: nmlrg [nmlrg-bounces@irtf.org] on behalf of Sheng Jiang [
>>> jiangsheng@huawei.com]
>>> Sent: 28 June 2016 21:03
>>> To: nmlrg@irtf.org
>>> Subject: [Nmlrg] slides of NMLRG #3, June 27th, Athens
>>>
>>> Hi, nmlrg,
>>>
>>> All slides that have been presented in our NMLRG #3 meeting, June 27th,
>>> 2016, Athens, Greece, with EUCNC2016, have been uploaded. They can be
>>> accessed through below link
>>>
>>>
>>> https://www.ietf.org/proceedings/interim/2016/06/27/nmlrg/proceedings.html
>>>
>>> Best regards,
>>>
>>> Sheng
>>> _______________________________________________
>>> nmlrg mailing list
>>> nmlrg@irtf.org
>>> https://www.irtf.org/mailman/listinfo/nmlrg
>>> _______________________________________________
>>> nmlrg mailing list
>>> nmlrg@irtf.org
>>> https://www.irtf.org/mailman/listinfo/nmlrg
>>>
>>
>>
>

--001a1140958411a5e30536850d34
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr">Hey Sheng, see in-line. Thx, --dmm<br><div class=3D"gmail_=
extra"><br><div class=3D"gmail_quote">On Thu, Jun 30, 2016 at 1:16 PM, Shen=
g Jiang <span dir=3D"ltr">&lt;<a href=3D"mailto:jiangsheng@huawei.com" targ=
et=3D"_blank">jiangsheng@huawei.com</a>&gt;</span> wrote:<br><blockquote cl=
ass=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-left-width:1px=
;border-left-style:solid;border-left-color:rgb(204,204,204);padding-left:1e=
x">




<div>
<div style=3D"direction:ltr;font-family:Tahoma;color:rgb(0,0,0);font-size:1=
0pt">
<p>Hi, David,</p>
<p>=C2=A0</p>
<p>Please see my replies with &lt;Sheng/&gt; in the lines.=C2=A0Sorry for b=
eing slow response due to my traveling.</p>
<p>=C2=A0</p>
<div style=3D"font-family:&quot;Times New Roman&quot;;color:rgb(0,0,0);font=
-size:16px">
<div>
<div dir=3D"ltr">
<div class=3D"gmail_extra">
<div class=3D"gmail_quote"><span class=3D"">
<blockquote style=3D"border-left-width:1px;border-left-style:solid;border-l=
eft-color:rgb(204,204,204);margin:0px 0px 0px 0.8ex;padding-left:1ex" class=
=3D"gmail_quote">
<div>
<div style=3D"font-family:Tahoma;direction:ltr;color:rgb(0,0,0);font-size:1=
0pt">
<p>However, upto now, I have serveral concerns. Firstly, before we could di=
scuss such dataset, we need to narrow our target scenarios to a few very=C2=
=A0specific use cases, giving that there are so many various scenarios in t=
he network area and they are so different
 from each other and require different dataset. </p>
</div>
</div>
</blockquote>
<div><br>
</div>
<div>I&#39;m hoping that not every network is not a one off. Rather, what w=
e seek is to learn what are the fundamental properties of networks that can=
 be the basis of reasoning about networks do. For example, we should be abl=
e to learn something from work done
 on transfer learning [0] and representation learning [1]. In the image rec=
ognition space there are even startups that provide pretrained AlexNet, Goo=
LeNet, etc trained on ImageNet (and others) that you can further train/cust=
omize for your particular application.
 In any event, if every network requires something totally different then w=
e haven&#39;t yet understood what is fundamental to the structure of our ne=
tworks.</div>
<div>=C2=A0</div>
</span><div>&lt;Sheng&gt;If I understand you correctly, you meant we should=
 do some generic learning in order to better understand the network fundame=
ntally, and the result should be helpful for solving these specific tasks i=
n the networks. It is different to what we have
 done - working on applying ML to solve specific network tasks. I would=C2=
=A0agree this proposed research direction are important and=C2=A0 I am very=
 interesting in such direction although=C2=A0 it looks ambitions and no gua=
rantee for success so far. &lt;/Sheng&gt;</div></div></div></div></div></di=
v></div></div></blockquote><div><br></div><div>No, not really. What I meant=
 is if every network is a one-off then by definition we haven&#39;t learned=
 to generalize. Now, generalization might not be possible, but I don&#39;t =
see any reason why it wouldn&#39;t be. It certainly is for perceptual tasks=
, so there&#39;s one data point (for example, if you want to do image recog=
nition you use a Conv Net because there is both theory and practice behind =
the framework).</div><div><br></div><blockquote class=3D"gmail_quote" style=
=3D"margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-style:solid;=
border-left-color:rgb(204,204,204);padding-left:1ex"><div><div style=3D"dir=
ection:ltr;font-family:Tahoma;color:rgb(0,0,0);font-size:10pt"><div style=
=3D"font-family:&quot;Times New Roman&quot;;color:rgb(0,0,0);font-size:16px=
"><div><div dir=3D"ltr"><div class=3D"gmail_extra"><div class=3D"gmail_quot=
e"><span class=3D"">
<div>=C2=A0</div>
<blockquote style=3D"border-left-width:1px;border-left-style:solid;border-l=
eft-color:rgb(204,204,204);margin:0px 0px 0px 0.8ex;padding-left:1ex" class=
=3D"gmail_quote">
<div>
<div style=3D"font-family:Tahoma;direction:ltr;color:rgb(0,0,0);font-size:1=
0pt">
<p>Secondly, as we did talk with operators, the real data is the important =
properties for them. So, it would be almost impossible for them to share th=
em. It leaves two possibilities on how to get dataset: a) generating datase=
t using simulation;</p>
</div>
</div>
</blockquote>
<div><br>
</div>
<div>This likely won&#39;t work. For example, if you do this then try to tr=
ain a DNN on that data set the NN will learn how the simulator works. This =
is unlikely to be the outcome that we&#39;d like (we already know how the s=
imulator works).</div>
<div><br>
</div>
<div>=C2=A0<span style=3D"font-family:Tahoma;color:rgb(0,0,0);font-size:10p=
t">b) try to get some partial dataset from operators that have removed thei=
r sensenble data. Both way are problemstic, in the risk that a lot of effor=
ts may waste in wrong areas.=C2=A0</span></div>
<div><span style=3D"font-family:Tahoma;color:rgb(0,0,0);font-size:10pt"><br=
>
</span></div>
<div><span style=3D"font-family:Tahoma;color:rgb(0,0,0);font-size:10pt">Agr=
ee that this is problematic.=C2=A0</span></div>
<span style=3D"font-family:Tahoma;color:rgb(0,0,0);font-size:10pt"></span><=
/span></div>
<span style=3D"font-family:Tahoma;color:rgb(0,0,0);font-size:10pt"></span><=
/div>
<span style=3D"font-family:Tahoma;color:rgb(0,0,0);font-size:10pt"></span><=
/div>
<span style=3D"font-family:Tahoma;color:rgb(0,0,0);font-size:10pt"></span><=
/div>
<div><span style=3D"font-family:Tahoma;color:rgb(0,0,0);font-size:10pt">
<div dir=3D"ltr" class=3D"gmail_quote">=C2=A0</div>
<div dir=3D"ltr" class=3D"gmail_quote">&lt;Sheng&gt;Here, we seems reach th=
e agreement on that we do NOT=C2=A0have a good way to obtain the useful dat=
aset. It is not good at all. Does that mean we have to wait for operators t=
o work on themselves alone? Any suggestion on how
 we could get quality real dataset?&lt;/Sheng&gt;<br></div></span></div></d=
iv></div></div></blockquote><div><br></div><div>This is a huge challenge, a=
nd one we should spend time talking about. Without something like ImageNet =
(or MNIST), we want to able to measure progress (or even if our techniques =
work).<br></div><div><br></div><div><br></div><blockquote class=3D"gmail_qu=
ote" style=3D"margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-st=
yle:solid;border-left-color:rgb(204,204,204);padding-left:1ex"><div><div st=
yle=3D"direction:ltr;font-family:Tahoma;color:rgb(0,0,0);font-size:10pt"><d=
iv style=3D"font-family:&quot;Times New Roman&quot;;color:rgb(0,0,0);font-s=
ize:16px"><div><span style=3D"font-family:Tahoma;color:rgb(0,0,0);font-size=
:10pt"><div dir=3D"ltr" class=3D"gmail_quote">
</div>
</span><span class=3D"">
<div dir=3D"ltr" class=3D"gmail_quote"><span style=3D"font-family:Tahoma;co=
lor:rgb(0,0,0);font-size:10pt">Thirdly, the availability of data=C2=A0may v=
ery different in different network environments. It depends on the measurem=
ent/congnitive functions that have been
 deployed or implemented. =C2=A0=C2=A0</span></div>
<div dir=3D"ltr" class=3D"gmail_quote"><span style=3D"font-family:Tahoma;co=
lor:rgb(0,0,0);font-size:10pt"><br>
</span></div>
<div dir=3D"ltr" class=3D"gmail_quote"><span style=3D"font-family:Tahoma;co=
lor:rgb(0,0,0);font-size:10pt">What is a &quot;cognitive function&quot;?</s=
pan></div>
<div dir=3D"ltr" class=3D"gmail_quote"><span style=3D"font-family:Tahoma;co=
lor:rgb(0,0,0);font-size:10pt"></span>=C2=A0</div>
</span><div dir=3D"ltr" class=3D"gmail_quote"><span style=3D"font-family:Ta=
homa;color:rgb(0,0,0);font-size:10pt">&lt;Sheng&gt;Here, I use the &quot;co=
gnitive function&quot; to represent that device or network function that=C2=
=A0collects the existing status or=C2=A0configuration data, against
 measurement.&lt;/Sheng&gt;</span></div></div></div></div></div></blockquot=
e><div><br></div><div><br></div><div>So is a network element, say a router,=
 a cognitive function?=C2=A0</div><div><br></div><blockquote class=3D"gmail=
_quote" style=3D"margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left=
-style:solid;border-left-color:rgb(204,204,204);padding-left:1ex"><div><div=
 style=3D"direction:ltr;font-family:Tahoma;color:rgb(0,0,0);font-size:10pt"=
><div style=3D"font-family:&quot;Times New Roman&quot;;color:rgb(0,0,0);fon=
t-size:16px"><div><span class=3D"">
<div dir=3D"ltr" class=3D"gmail_quote"><span style=3D"font-family:Tahoma;co=
lor:rgb(0,0,0);font-size:10pt"><br>
</span></div>
<div dir=3D"ltr" class=3D"gmail_quote"><span style=3D"font-family:Tahoma;co=
lor:rgb(0,0,0);font-size:10pt">In any event, you have outlined just a few o=
f the problems with creating &quot;standardized&quot; data sets for our use=
. I have quoted standardized because I&#39;m beginning
 to think its the wrong word (invokes &quot;standards&quot; e.g., IETF, &lt=
;X&gt;..., and that is not what is meant here).=C2=A0 Not sure what the rig=
ht term is but =C2=A0&quot;standardized&quot; seems to be the wrong word he=
re.</span></div>
<div dir=3D"ltr" class=3D"gmail_quote"><span style=3D"font-family:Tahoma;co=
lor:rgb(0,0,0);font-size:10pt"><br>
</span></div>
<blockquote style=3D"border-left-width:1px;border-left-style:solid;border-l=
eft-color:rgb(204,204,204);margin:0px 0px 0px 0.8ex;padding-left:1ex" dir=
=3D"ltr" class=3D"gmail_quote">
<div style=3D"font-family:Tahoma;direction:ltr;color:rgb(0,0,0);font-size:1=
0pt">
<p>=C2=A0</p>
<p>Although having the above concerns, I =C2=A0think we should: a) research=
 on certain dataset. It=C2=A0would be very useful to prove the machine lear=
ning could perform better or provide more flexibility and adaptibility;
</p>
</div>
</blockquote>
<div dir=3D"ltr" class=3D"gmail_quote"><br>
</div>
<div dir=3D"ltr" class=3D"gmail_quote">Not sure what you mean here. By &quo=
t;certain dataset&quot; you mean what?</div>
<div dir=3D"ltr" class=3D"gmail_quote">=C2=A0</div>
</span><div dir=3D"ltr" class=3D"gmail_quote">&lt;Sheng&gt;We did discuss a=
 potential standardized dataset, didn&#39;t we?&lt;/Sheng&gt;</div></div></=
div></div></div></blockquote><div><br></div><div>Sorry, =C2=A0I didn&#39;t =
understand the language.</div><div>=C2=A0</div><blockquote class=3D"gmail_q=
uote" style=3D"margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-s=
tyle:solid;border-left-color:rgb(204,204,204);padding-left:1ex"><div><div s=
tyle=3D"direction:ltr;font-family:Tahoma;color:rgb(0,0,0);font-size:10pt"><=
div style=3D"font-family:&quot;Times New Roman&quot;;color:rgb(0,0,0);font-=
size:16px"><div><span class=3D"">
<div dir=3D"ltr" class=3D"gmail_quote">=C2=A0</div>
<blockquote style=3D"border-left-width:1px;border-left-style:solid;border-l=
eft-color:rgb(204,204,204);margin:0px 0px 0px 0.8ex;padding-left:1ex" dir=
=3D"ltr" class=3D"gmail_quote">
<div style=3D"font-family:Tahoma;direction:ltr;color:rgb(0,0,0);font-size:1=
0pt">
<p>b) continue our discuss on various network use cases that could benefit =
from applying machine learning, in order to serve wider network communities=
.</p>
</div>
</blockquote>
<div dir=3D"ltr" class=3D"gmail_quote"><br>
</div>
<div dir=3D"ltr" class=3D"gmail_quote">Use cases are good, however machine =
learning isn&#39;t something we can just apply to a use case because we&#39=
;ve defined the use case. There are really at least three parts of this tha=
t need to be considered as part of a solution:
 (i). use case (as you mention), (ii). algorithm(s) that are going to provi=
de the regression/classification/ desired in (i), and (iii). data sets.=C2=
=A0</div>
<div dir=3D"ltr" class=3D"gmail_quote"><br>
</div>
<div dir=3D"ltr" class=3D"gmail_quote">(ii). requires understanding of the =
algorithms that can be applied to produce the inferences (regressions, ...)=
 =C2=A0we are interested in. What algorithms we can use depend on the data =
we can get (iii). =C2=A0</div>
<div dir=3D"ltr" class=3D"gmail_quote">=C2=A0</div>
</span><div dir=3D"ltr" class=3D"gmail_quote">&lt;Sheng&gt;What we have don=
e up to now is, in a specific use case, to design an algorithm or a combina=
tion of algorithm on chosen dataset (this may be dynamic data from the netw=
ork environment) to complete a specific task.&lt;/Sheng&gt;</div>
</div>
<div><span class=3D"">
<div dir=3D"ltr" class=3D"gmail_quote">=C2=A0</div>
<div dir=3D"ltr" class=3D"gmail_quote">That said, one of the things we&#39;=
ve learned over the past decade or so is that learning itself is a very gen=
eral process. We see this in how powerful something as simple as SGD/backpr=
op is (see e.g., [2]). We can also use this
 to our advantage and re-purpose algorithms originally developed for other =
tasks. A simple example of this is LDA (see [3,4]). Here the original idea =
was to model topics in a corpus of documents, where the topics themselves (=
along with the per-document topic
 proportions) were the latent/hidden variables. One way to re-purpose LDA f=
or applications like ours (essentially classification/anomaly detection tas=
ks) is to view each to/from pair of IP addresses as a &quot;document&quot; =
(in the language of LDA) and then words are
 constructed from features of the to/from traffic (e.g., to ports, etc). Se=
e [5] for a decent example of how you might think about this kind of &quot;=
statistical clustering&quot; using the LDA Probabilistic Graphical Model ap=
proach. So there is hope, but there is a lot
 of work left to be done on both the theory and practice sides of this (whi=
ch as I mentioned earlier are, at least currently, essentially the same thi=
ng).</div>
<div dir=3D"ltr" class=3D"gmail_quote">=C2=A0</div>
</span><div dir=3D"ltr" class=3D"gmail_quote">&lt;Sheng&gt;Agree on that le=
arning and learning algorithm is general. But what we did was only use it i=
n specific way to solve specific tasks/use cases. </div></div></div></div><=
/div></blockquote><div><br></div><div>Understand, but I&#39;m questioning w=
hether this is a legitimate topic for a research group. Building one-off so=
lutions is obviously possible (look at all of the startups, products etc in=
 this space), but none of these approachs give us insight into how to learn=
 from network data, what its properties are, or how to generalize (really t=
he holy grail).=C2=A0</div><div><br></div><blockquote class=3D"gmail_quote"=
 style=3D"margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-style:=
solid;border-left-color:rgb(204,204,204);padding-left:1ex"><div><div style=
=3D"direction:ltr;font-family:Tahoma;color:rgb(0,0,0);font-size:10pt"><div =
style=3D"font-family:&quot;Times New Roman&quot;;color:rgb(0,0,0);font-size=
:16px"><div><div dir=3D"ltr" class=3D"gmail_quote">As I agreed earlier, gen=
erlized learning on network could be an important
 direction and I am interesting in. Let&#39;s see how far we could go. On a=
nother side, I have to emphasis that the general learning must be helpful o=
n solving specific tasks, in other words, specific tasks should be earlier =
to solve from the result of general
 learning. Otherwise, the general learning is not mearningful.&lt;/Sheng&gt=
;</div><span class=3D"">
<div dir=3D"ltr" class=3D"gmail_quote">=C2=A0</div>
<blockquote style=3D"border-left-width:1px;border-left-style:solid;border-l=
eft-color:rgb(204,204,204);margin:0px 0px 0px 0.8ex;padding-left:1ex" dir=
=3D"ltr" class=3D"gmail_quote">
<div style=3D"font-family:Tahoma;direction:ltr;color:rgb(0,0,0);font-size:1=
0pt">
<p>Another important point I would like to discuss is that you mentioned &q=
uot;our goal is to provide accurate, repeatable, and explainable results.&q=
uot; Ideally, we do want this. It is our traditional way for logic-based pr=
ogramming.</p>
</div>
</blockquote>
<div dir=3D"ltr" class=3D"gmail_quote">What is &quot;logic-based programmin=
g&quot;?=C2=A0 Do you mean logic programming, e.g., Prolog? If so I underst=
and logic programming (I did a lot of work on logic programming back in the=
 late 80s/early 90s, see=C2=A0<a href=3D"http://dl.acm.org/citation.cfm?id=
=3D87991" target=3D"_blank">http://dl.acm.org/citation.cfm?id=3D87991</a>).
 But even so, I don&#39;t understand the connection, so can you be more exp=
licit?</div>
<div dir=3D"ltr" class=3D"gmail_quote">=C2=A0</div>
</span><div dir=3D"ltr" class=3D"gmail_quote">&lt;Sheng&gt;Here, &quot;logi=
c-based programming&quot; I referred to all traditional programming that re=
alize human pre-designed logic into computing programs. It is comparing to =
ML-based programming.&lt;/Sheng&gt;</div></div></div></div></div></blockquo=
te><div><br></div><div>None of those terms are well defined, but I get your=
 sense.=C2=A0</div><blockquote class=3D"gmail_quote" style=3D"margin:0px 0p=
x 0px 0.8ex;border-left-width:1px;border-left-style:solid;border-left-color=
:rgb(204,204,204);padding-left:1ex"><div><div style=3D"direction:ltr;font-f=
amily:Tahoma;color:rgb(0,0,0);font-size:10pt"><div style=3D"font-family:&qu=
ot;Times New Roman&quot;;color:rgb(0,0,0);font-size:16px"><div><span class=
=3D"">
<div dir=3D"ltr" class=3D"gmail_quote">=C2=A0</div>
<blockquote style=3D"border-left-width:1px;border-left-style:solid;border-l=
eft-color:rgb(204,204,204);margin:0px 0px 0px 0.8ex;padding-left:1ex" dir=
=3D"ltr" class=3D"gmail_quote">
<div style=3D"font-family:Tahoma;direction:ltr;color:rgb(0,0,0);font-size:1=
0pt">
<p>However, I am not sure that is a doable target=C2=A0with machine learnin=
g. Many of machine learning algorithms=C2=A0are based on statistics theory,=
 which may not be that much accurate.
</p>
</div>
</blockquote>
<div dir=3D"ltr" class=3D"gmail_quote">Sorry, I don&#39;t understand this. =
Things like Bayesian inference (e.g., LDA) are inherently about quantifying=
 uncertainty. In any event can you be more specific? A learning algorithm c=
an &quot;accurately&quot; approximate something like
 a posterior distribution, its just that the output is just that, a posteri=
or distribution. So there is a difference between &quot;accuracy&quot; (whi=
ch isn&#39;t really defined here) and the stochastic nature of the results =
we get from ML.</div>
<div dir=3D"ltr" class=3D"gmail_quote">=C2=A0</div>
<div dir=3D"ltr" class=3D"gmail_quote">OTOH, if you mean accuracy of our es=
timated distributions then there are obviously ways this can go wrong. For =
example, even an algorithm as simple as k-means can go wrong since the opti=
mization objective (distortion function)
 is non-convex, so it is sensitive to initial conditions (in particular, wh=
ere you start; if you are unlucky you can get stuck in a local minimum =C2=
=A0due to the non-convex nature of the objective). See e.g., [8]. BTW, =C2=
=A0k-means was called out in one of the decks
 below but as I mentioned, not much discussion of the important topics we a=
re discussing here there. There is also approximate inference (for example,=
 MCMC or variational approaches), but these approaches also have provably=
=C2=A0quantitative error bounds (<a href=3D"http://www.1-4-5.net/~dmm/ml/ps=
.pdf" target=3D"_blank">http://www.1-4-5.net/~dmm/ml/ps.pdf</a>
 talks a bit about how variational approaches, essential trading inference =
for optimization where the inference (integral) is intractable, find a tigh=
t lower-bound using Jensen&#39;s inequality.=C2=A0</div>
<div dir=3D"ltr" class=3D"gmail_quote">=C2=A0</div>
</span><div dir=3D"ltr" class=3D"gmail_quote">
<div dir=3D"ltr" class=3D"gmail_quote">&lt;Sheng&gt;The reason that statist=
ics-based ML algorithms can work is that they chose to focus on only high p=
robability and igrone corner cases. </div></div></div></div></div></div></b=
lockquote><div><br></div><div>I don&#39;t believe this is correct, but perh=
aps I don&#39;t understand your point. Can you give a concrete example of (=
a). what you mean by statistics-based ML algorithm and (b). one that ignore=
s corner cases and how/why?</div><div>=C2=A0</div><blockquote class=3D"gmai=
l_quote" style=3D"margin:0px 0px 0px 0.8ex;border-left-width:1px;border-lef=
t-style:solid;border-left-color:rgb(204,204,204);padding-left:1ex"><div><di=
v style=3D"direction:ltr;font-family:Tahoma;color:rgb(0,0,0);font-size:10pt=
"><div style=3D"font-family:&quot;Times New Roman&quot;;color:rgb(0,0,0);fo=
nt-size:16px"><div><div dir=3D"ltr" class=3D"gmail_quote"><div dir=3D"ltr" =
class=3D"gmail_quote">But when mearningful corner cases happen or we want t=
o target the corner cases, machine
 learning may not accurately work. Again, &quot;accuracy&quot; itself does =
not properly defined. So, we may talk about very different things. On other=
 side, I also agree that accurate is not important topic here. ML can reach=
 a high-level accurate, which are suffient
 for majority cases. But ML may never reach full accurate while traditional=
 program can 100% certain the result on giving circumstance.&lt;/Sheng&gt;<=
/div>
<div dir=3D"ltr" class=3D"gmail_quote">=C2=A0</div>
</div><span class=3D"">
<blockquote style=3D"border-left-width:1px;border-left-style:solid;border-l=
eft-color:rgb(204,204,204);margin:0px 0px 0px 0.8ex;padding-left:1ex" dir=
=3D"ltr" class=3D"gmail_quote">
<div style=3D"font-family:Tahoma;direction:ltr;color:rgb(0,0,0);font-size:1=
0pt">
<p>Some of machine learning algorithms=C2=A0are black box or have very comp=
lexicated internal computing process. So, the resulf of these algorithms=C2=
=A0may be hard to be explained.
</p>
</div>
</blockquote>
<div dir=3D"ltr" class=3D"gmail_quote">This is definitely true to some exte=
nt today with DNNs, but that is changing rapidly. Results can be repeatable=
 but explaining exactly what units (artificial neurons) far from the input =
are doing can be difficult. But this
 too is an area of active research and development (see [6] from last year&=
#39;s ICML, or the work I pointed to on GANs in my last note as progress in=
 this area). Other approaches such probabilistic graphical models or signal=
 processing approaches are perhaps more
 easily explained, but I don&#39;t think this is the level of explanation t=
hat is going to be useful to people deploying and operating networks, so we=
&#39;ll have to find different ways of explaining what things like DNNs to =
in a way that is consumable by operators,
 etc.</div>
<div dir=3D"ltr" class=3D"gmail_quote">=C2=A0</div>
</span><div dir=3D"ltr" class=3D"gmail_quote">&lt;Sheng&gt;Sharing the same=
 opinion here. Every we could really explain how DNNs works in a very compl=
icated way. Our customers would not have enough patient to listen. They wou=
ld still not trust ML. We, actually the whole AI/ML
 society, needs to find other way to explain and prove things like DNN are =
trustable. Only then, they can be use for more serious or autonomic=C2=A0ta=
sks.&lt;/Sheng&gt;</div></div></div></div></div></blockquote><div><br></div=
><div>I&#39;m optimistic that we will be able to understand what the higher=
 layers in DNNs are doing; much progress has already been and is being made=
. That said, this problem is really more of an issue for consumers of ML te=
chnology, and not so much for those building it (although it would be usefu=
l to better understand the computation a DNN is making; as I mentioned much=
 progress has been made here). As a loose analogy, how many consumers of th=
e Linux Kernel really understand what it is doing in any detailed way? Of c=
ourse explainability is important and I don&#39;t mean to diminish that imp=
ortance in any way.</div><div><br></div><div>BTW, people are going to deplo=
y ML based systems not because they are interesting or cool (which they are=
), but rather because their competitors are/will. We already see this in th=
e hyper-scales, automotive industry, process automation industry, and even =
our industry (check out Chris&#39; description of ML as &quot;automating yo=
ur automation&quot; =C2=A0<a href=3D"https://www.youtube.com/watch?v=3DPZAA=
OVbiK9A">https://www.youtube.com/watch?v=3DPZAAOVbiK9A</a>).=C2=A0</div><di=
v><br></div><blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0=
.8ex;border-left-width:1px;border-left-style:solid;border-left-color:rgb(20=
4,204,204);padding-left:1ex"><div><div style=3D"direction:ltr;font-family:T=
ahoma;color:rgb(0,0,0);font-size:10pt"><div style=3D"font-family:&quot;Time=
s New Roman&quot;;color:rgb(0,0,0);font-size:16px"><div><span class=3D""><d=
iv dir=3D"ltr" class=3D"gmail_quote"><br>
</div>
<div dir=3D"ltr" class=3D"gmail_quote">All this by way of saying that we ar=
e gaining better understanding how DNNs work every day (literally) and as a=
 result =C2=A0I&#39;m optimistic that we can build technology that will be =
accurate (again, stochastic results), repeatable
 and explainable.</div>
<div dir=3D"ltr" class=3D"gmail_quote">=C2=A0</div>
</span><div dir=3D"ltr" class=3D"gmail_quote">&lt;Sheng&gt;I hope I can sha=
re you optimistic. Actually, I do believe we could do so one day. But I am =
not sure how long it will take. &lt;/Sheng&gt;</div><span class=3D"">
<div dir=3D"ltr" class=3D"gmail_quote"><br>
</div>
<div dir=3D"ltr" class=3D"gmail_quote">I&#39;ll just note here that if you =
have a model with literally billions of parameters being trained by SGD its=
 not that surprising that humans have a hard time understanding what the co=
mputation is. As a concrete example, DNNs
 can solve the so-called &quot;selectivity-invariance&quot; problem [7], bu=
t we humans don&#39;t really =C2=A0know how to do, so this is another examp=
le.=C2=A0</div>
<div dir=3D"ltr" class=3D"gmail_quote">=C2=A0</div>
<blockquote style=3D"border-left-width:1px;border-left-style:solid;border-l=
eft-color:rgb(204,204,204);margin:0px 0px 0px 0.8ex;padding-left:1ex" dir=
=3D"ltr" class=3D"gmail_quote">
<div style=3D"font-family:Tahoma;direction:ltr;color:rgb(0,0,0);font-size:1=
0pt">
<p>Also the process of machine learning may be hard to be controlled or int=
ervened.
</p>
</div>
</blockquote>
<div dir=3D"ltr" class=3D"gmail_quote"><br>
</div>
<div dir=3D"ltr" class=3D"gmail_quote">Again, not sure what you mean here. =
Please give a bit more detail.</div>
<div dir=3D"ltr" class=3D"gmail_quote">=C2=A0</div>
</span><div dir=3D"ltr" class=3D"gmail_quote">&lt;Sheng&gt; Here, I actuall=
y refer to DNN too. We don&#39;t know how to intervened DNN in its running =
time.</div><span class=3D"">
<div dir=3D"ltr" class=3D"gmail_quote">=C2=A0</div></span></div></div></div=
></div></blockquote><div><br></div><div>Do you mean that once you&#39;ve tr=
ained a DNN you can only use it for classification (or whatever) and can&#3=
9;t further train it, or do you mean that we can&#39;t learn from additiona=
l data, or something else?</div><div>=C2=A0</div><blockquote class=3D"gmail=
_quote" style=3D"margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left=
-style:solid;border-left-color:rgb(204,204,204);padding-left:1ex"><div><div=
 style=3D"direction:ltr;font-family:Tahoma;color:rgb(0,0,0);font-size:10pt"=
><div style=3D"font-family:&quot;Times New Roman&quot;;color:rgb(0,0,0);fon=
t-size:16px"><div><span class=3D"">
<blockquote style=3D"border-left-width:1px;border-left-style:solid;border-l=
eft-color:rgb(204,204,204);margin:0px 0px 0px 0.8ex;padding-left:1ex" dir=
=3D"ltr" class=3D"gmail_quote">
<div style=3D"font-family:Tahoma;direction:ltr;color:rgb(0,0,0);font-size:1=
0pt">
<p>Repeatable, yes, we have to have this feature to be widely adopted. Howe=
ver, repeatable may have to come with some level of toleration for the inac=
curate.</p>
</div>
</blockquote>
<div dir=3D"ltr" class=3D"gmail_quote"><br>
</div>
<div dir=3D"ltr" class=3D"gmail_quote">All of this is inherently probabilis=
tic. A DNN, for example, can output the posterior distribution over some hi=
dden variables of interest accurately. LDA finds the latent &quot;topics&qu=
ot; and word distributions, both of which are
 multinomals so again we&#39;re essentially getting a posterior over the hi=
dden variables.=C2=A0 Even k-means is =C2=A0just the EM (Expectation Maximi=
zation) algorithm applied to a particular Naive Bayes model. The same can b=
e said of almost every learning algorithm you
 might mention. So &quot;inaccurate&quot; isn&#39;t the correct word here (=
unless you mean whatever you trained didn&#39;t predict whatever you wanted=
 to predict, in which case something else is wrong). So the DNN gives you a=
s stochastic result, which may be different than what
 many network people are used to. But its not inaccurate, its just stochast=
ic.</div>
<div dir=3D"ltr" class=3D"gmail_quote">=C2=A0</div>
</span><div dir=3D"ltr" class=3D"gmail_quote">&lt;Sheng&gt;Inaccurate or st=
ochastic, they are not what many network people can accept. In their eyes, =
network is definite, every parameter is definite. However, actually, if you=
 could analyze every network parameter, the value
 may be definite, but actually they all have toleration for some level of i=
naccurate as long as their design purposes could meet. It is the fundementa=
l that we could use ML to solve network task.</div></div></div></div></div>=
</blockquote><div><br></div><div><br></div><div>I didn&#39;t understand wha=
t you mean here.=C2=A0</div><div><br></div><div>Thanks,</div><div><br></div=
><div>--dmm</div><div>=C2=A0</div><blockquote class=3D"gmail_quote" style=
=3D"margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-style:solid;=
border-left-color:rgb(204,204,204);padding-left:1ex"><div><div style=3D"dir=
ection:ltr;font-family:Tahoma;color:rgb(0,0,0);font-size:10pt"><div style=
=3D"font-family:&quot;Times New Roman&quot;;color:rgb(0,0,0);font-size:16px=
"><div>
<div dir=3D"ltr" class=3D"gmail_quote">=C2=A0</div>
<div dir=3D"ltr" class=3D"gmail_quote">Thanks so much. Regards,</div><span =
class=3D""><font color=3D"#888888">
<div dir=3D"ltr" class=3D"gmail_quote">=C2=A0</div>
<div dir=3D"ltr" class=3D"gmail_quote">Sheng &lt;/Sheng&gt;</div></font></s=
pan><div><div class=3D"h5">
<div dir=3D"ltr" class=3D"gmail_quote">=C2=A0</div>
<div dir=3D"ltr" class=3D"gmail_quote"><br>
=C2=A0</div>
<div dir=3D"ltr" class=3D"gmail_quote">Thx,</div>
<div dir=3D"ltr" class=3D"gmail_quote"><br>
</div>
<div dir=3D"ltr" class=3D"gmail_quote">--dmm</div>
<div dir=3D"ltr" class=3D"gmail_quote"><br>
</div>
<div dir=3D"ltr" class=3D"gmail_quote"><br>
</div>
<div dir=3D"ltr" class=3D"gmail_quote"><br>
</div>
<div dir=3D"ltr" class=3D"gmail_quote">[0]=C2=A0<a href=3D"http://people.cs=
ail.mit.edu/mcollins/papers/CVPR_2008.pdf" target=3D"_blank">http://people.=
csail.mit.edu/mcollins/papers/CVPR_2008.pdf</a></div>
<div dir=3D"ltr" class=3D"gmail_quote">[1] =C2=A0<span style=3D"font-family=
:Tahoma;color:rgb(0,0,0);font-size:10pt"><a href=3D"https://arxiv.org/abs/1=
206.5538" target=3D"_blank">https://arxiv.org/abs/1206.5538</a></span></div=
>
<div dir=3D"ltr" class=3D"gmail_quote"><span style=3D"font-family:Tahoma;co=
lor:rgb(0,0,0);font-size:10pt">[2]=C2=A0</span><font color=3D"#000000" face=
=3D"Tahoma"><span style=3D"font-size:13px"><a href=3D"http://deliprao.com/a=
rchives/153" target=3D"_blank">http://deliprao.com/archives/153</a></span><=
/font></div>
<div dir=3D"ltr" class=3D"gmail_quote"><font color=3D"#000000" face=3D"Taho=
ma"><span style=3D"font-size:13px">[3]=C2=A0<a href=3D"http://www.1-4-5.net=
/~dmm/ml/lda_intro.pdf" target=3D"_blank">http://www.1-4-5.net/~dmm/ml/lda_=
intro.pdf</a></span></font></div>
<div dir=3D"ltr" class=3D"gmail_quote"><font color=3D"#000000" face=3D"Taho=
ma"><span style=3D"font-size:13px">[4]=C2=A0<a href=3D"http://www.jmlr.org/=
papers/volume3/blei03a/blei03a.pdf" target=3D"_blank">http://www.jmlr.org/p=
apers/volume3/blei03a/blei03a.pdf</a></span></font></div>
<div dir=3D"ltr" class=3D"gmail_quote"><font color=3D"#000000" face=3D"Taho=
ma"><span style=3D"font-size:13px">[5]=C2=A0<a href=3D"http://ieeexplore.ie=
ee.org/stamp/stamp.jsp?arnumber=3D5949386" target=3D"_blank">http://ieeexpl=
ore.ieee.org/stamp/stamp.jsp?arnumber=3D5949386</a></span></font></div>
<div dir=3D"ltr" class=3D"gmail_quote"><font color=3D"#000000" face=3D"Taho=
ma"><span style=3D"font-size:13px">[6]=C2=A0<a href=3D"http://yosinski.com/=
media/papers/Yosinski__2015__ICML_DL__Understanding_Neural_Networks_Through=
_Deep_Visualization__.pdf" target=3D"_blank">http://yosinski.com/media/pape=
rs/Yosinski__2015__ICML_DL__Understanding_Neural_Networks_Through_Deep_Visu=
alization__.pdf</a></span></font></div>
<div dir=3D"ltr" class=3D"gmail_quote"><font color=3D"#000000" face=3D"Taho=
ma"><span style=3D"font-size:13px">[7]=C2=A0<a href=3D"https://cbmm.mit.edu=
/sites/default/files/publications/Anselmi_Invariance_CBMM_memo_29.pdf" targ=
et=3D"_blank">https://cbmm.mit.edu/sites/default/files/publications/Anselmi=
_Invariance_CBMM_memo_29.pdf</a></span></font></div>
<div dir=3D"ltr" class=3D"gmail_quote"><font color=3D"#000000" face=3D"Taho=
ma"><span style=3D"font-size:13px">[8]=C2=A0<a href=3D"http://cs229.stanfor=
d.edu/notes/cs229-notes7a.pdf" target=3D"_blank">http://cs229.stanford.edu/=
notes/cs229-notes7a.pdf</a></span></font></div>
<blockquote style=3D"border-left-width:1px;border-left-style:solid;border-l=
eft-color:rgb(204,204,204);margin:0px 0px 0px 0.8ex;padding-left:1ex" dir=
=3D"ltr" class=3D"gmail_quote">
<div>
<div style=3D"direction:ltr">
<p style=3D"font-family:Tahoma;color:rgb(0,0,0);font-size:10pt">Best regard=
s,</p>
<p style=3D"font-family:Tahoma;color:rgb(0,0,0);font-size:10pt">=C2=A0</p>
<p style=3D"font-family:Tahoma;color:rgb(0,0,0);font-size:10pt">Sheng</p>
<p style=3D"font-family:Tahoma;color:rgb(0,0,0);font-size:10pt">=C2=A0</p>
<p style=3D"font-family:Tahoma;color:rgb(0,0,0);font-size:10pt"></p>
<hr style=3D"font-family:Tahoma;color:rgb(0,0,0);font-size:10pt">
<p style=3D"font-family:Tahoma;color:rgb(0,0,0);font-size:10pt"></p>
<div>
<div style=3D"direction:ltr"><font color=3D"#000000" size=3D"2" face=3D"Tah=
oma"><b style=3D"font-family:&quot;Times New Roman&quot;;color:rgb(0,0,0);f=
ont-size:16px">From:</b><font face=3D"Times New Roman"><span style=3D"font-=
size:16px"> David Meyer [</span></font><a style=3D"font-family:&quot;Times =
New Roman&quot;;color:rgb(0,0,0);font-size:16px" href=3D"mailto:dmm@1-4-5.n=
et" target=3D"_blank">dmm@1-4-5.net</a><font face=3D"Times New Roman"><span=
 style=3D"font-size:16px">]</span></font><br>
<b style=3D"font-family:&quot;Times New Roman&quot;;color:rgb(0,0,0);font-s=
ize:16px">Sent:</b><font face=3D"Times New Roman"><span style=3D"font-size:=
16px"> 28 June 2016 23:30</span></font><a href=3D"http://deliprao.com/archi=
ves/153" target=3D"_blank">http://deliprao.com/archives/153</a><br>
<b style=3D"font-family:&quot;Times New Roman&quot;;color:rgb(0,0,0);font-s=
ize:16px">To:</b><font face=3D"Times New Roman"><span style=3D"font-size:16=
px"> Sheng Jiang</span></font><br>
<b style=3D"font-family:&quot;Times New Roman&quot;;color:rgb(0,0,0);font-s=
ize:16px">Cc:</b>
<a style=3D"font-family:&quot;Times New Roman&quot;;color:rgb(0,0,0);font-s=
ize:16px" href=3D"mailto:nmlrg@irtf.org" target=3D"_blank">
nmlrg@irtf.org</a><br>
<b style=3D"font-family:&quot;Times New Roman&quot;;color:rgb(0,0,0);font-s=
ize:16px">Subject:</b><font face=3D"Times New Roman"><span style=3D"font-si=
ze:16px"> Re: [Nmlrg] links to each slide of presentations//RE: slides of N=
MLRG #3, June 27th, Athens</span></font><br>
</font><br>
</div>
<div style=3D"font-family:&quot;Times New Roman&quot;;color:rgb(0,0,0);font=
-size:16px">
<div>
<div></div>
<div>
<div dir=3D"ltr">Sheng,
<div><br>
</div>
<div>Thanks for the pointers. One thing I notice is that we don&#39;t have =
much on what the characteristics of network data might be, and as such, wha=
t kinds of existing learning algorithms might be suitable and where additio=
nal development might be required. For
 example, is the collected data IID? Is it time series? Is the underlying d=
istribution stationary? If not, how does that constrain algorithms we might=
 use or develop? For example, how does a given algorithm deal with concept =
drift/internal covariate shift (in
 the case of DNNs; see e.g., batch normalization [0]). There are many other=
 such questions, such as is the data categorical (e.g., ports, IP addresses=
) or is it=C2=A0continuous/discrete (e.g., counters). And if the data in qu=
estion is categorical, what is the=C2=A0cardinality
 of the categories (this will inform how such data can be encoded); in the =
case of IP addresses we can&#39;t really one-hot encode addresses because t=
heir cardinality is too large (2^32 or 2^128); this has implications for ho=
w we build classifiers (in particular,
 for softmax layers in DNNs of various kinds).</div>
<div><br>
</div>
<div>Related to the above is the question of features? What are good featur=
es for networking? Where do they come from? Are they domain specific? Can w=
e learn features the way a DNNs does in the network space? Can we use autoe=
ncoders to discover such features?
 Or can we use GANs to train DNNs for network classification tasks in an un=
supervised manner? Are there other, non-ad-hoc (well founded) methods we ca=
n use, or is every use case a one-off (one would hope not).</div>
<div><br>
</div>
<div>We can carry the same kinds of analyzes to the algorithms applied. For=
 example, while something k-means is an effective way to get a feeling for =
how=C2=A0continuous/discrete hangs together, if our data is categorical sta=
tistical clustering approaches such as
 LDA might provide a more well-founded approach (of course, as with most Ba=
yesian techniques is the question of approximate inference arises since in =
most interesting cases the integral that we need to solve, namely the margi=
nal probability data isn&#39;t tractable
 so we need to resort to MCMC or more likely=C2=A0variational inference). A=
nd what about the use of SGD/batch normalization etc with DNNs, and perhaps=
 more importantly, can we use network data to train DNN policy networks for=
 reinforcement learning like we saw in
 deep Q-learning and AlphaGo?</div>
<div><br>
</div>
<div>These comments are all by way of saying that we don&#39;t have a solid=
 theoretical understanding (yet) of how techniques that have been so succes=
sful in other domains (e.g., DNNs for perceptual tasks) generalize to netwo=
rking use cases. We will need this understanding
 if our goal is to provide accurate, repeatable, and explainable results.=
=C2=A0</div>
<div><br>
</div>
<div>In order to accomplish all of this we need, as I have been saying , no=
t only a good understanding of how these algorithms work but also standardi=
zed data sets and associated benchmarks so we can tell if we are making pro=
gress (or even if our techniques
 work). Analogies here include MNIST and ImageNet and their associated benc=
hmarks, among others. As mentioned =C2=A0standardized data sets are key to =
making progress in the ML for networking space (otherwise how do you know y=
our technique works and/or improves on
 another techniques?). One might assume that these data sets would need to =
be labeled (as supervised learning is where most of the progress is being m=
ade these days), but not necessarily; Generative Adversarial Networks (GANs=
) have emerged as a new way to train
 DNNs in an unsupervised manner (this is moving very rapidly; see e.g.,=C2=
=A0<a href=3D"https://openai.com/blog/generative-models/" target=3D"_blank"=
>https://openai.com/blog/generative-models/</a>).</div>
<div><br>
</div>
<div>The summary here is that the &quot;distance&quot; between theory and p=
ractice in ML is effectively zero right now due to the incredible rate of p=
rogress in the field; this means we need =C2=A0to understand both sides of =
the theory/practice coin in order to be effective.
 None of the slide decks provide much background on what the proposed algor=
ithms are, how they work, or why they should be expected to work on network=
 data.</div>
<div><br>
</div>
<div>Finally, if you are interested in LDA or other algorithms there are a =
few short explanatory pieces I have written for my team on
<a href=3D"http://www.1-4-5.net/~dmm/ml" target=3D"_blank">http://www.1-4-5=
.net/~dmm/ml</a> (works in progress).</div>
<div><br>
</div>
<div>Thanks,</div>
<div><br>
Dave</div>
<div><br>
</div>
<div>[0]=C2=A0<a href=3D"https://arxiv.org/pdf/1502.03167.pdf" target=3D"_b=
lank">https://arxiv.org/pdf/1502.03167.pdf</a></div>
</div>
<div class=3D"gmail_extra"><br>
<div class=3D"gmail_quote">On Tue, Jun 28, 2016 at 7:03 AM, Sheng Jiang <sp=
an dir=3D"ltr">
&lt;<a href=3D"mailto:jiangsheng@huawei.com" target=3D"_blank">jiangsheng@h=
uawei.com</a>&gt;</span> wrote:<br>
<blockquote style=3D"border-left-width:1px;border-left-style:solid;border-l=
eft-color:rgb(204,204,204);margin:0px 0px 0px 0.8ex;padding-left:1ex" class=
=3D"gmail_quote">
Oops... The proceeding page for interim meeting seems not be as intelligent=
 as the proceeding pages of IETF meetings. Our proceeding page does not aut=
onomically show the slides. I have sent an email to IETF secretary to ask t=
hem to fix it. Meanwhile, in this
 email, here is the links for each presentations:<br>
<br>
Slide Filename Edit Replace Delete<br>
Chair Slides<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-0.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-0.pdf</a><br>
<br>
Introduction to Network Machine Learning &amp; NMLRG<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-1.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-1.pdf</a><br>
<br>
Data Collection and Analysis At High Security Lab<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-2.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-2.pdf</a><br>
<br>
Use Cases of Applying Machine Learning Mechanism with Network Traffic<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-3.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-3.pdf</a><br>
<br>
Mobile network state characterization and prediction<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-4.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-4.pdf</a><br>
<br>
Learning how to route<br>
<a href=3D"https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/sl=
ides-interim-2016-nmlrg-01-5.pdf" rel=3D"noreferrer" target=3D"_blank">http=
s://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-20=
16-nmlrg-01-5.pdf</a><br>
<br>
Regards,<br>
<br>
Sheng<br>
________________________________________<br>
From: nmlrg [<a href=3D"mailto:nmlrg-bounces@irtf.org" target=3D"_blank">nm=
lrg-bounces@irtf.org</a>] on behalf of Sheng Jiang [<a href=3D"mailto:jiang=
sheng@huawei.com" target=3D"_blank">jiangsheng@huawei.com</a>]<br>
Sent: 28 June 2016 21:03<br>
To: <a href=3D"mailto:nmlrg@irtf.org" target=3D"_blank">nmlrg@irtf.org</a><=
br>
Subject: [Nmlrg] slides of NMLRG #3, June 27th, Athens<br>
<br>
Hi, nmlrg,<br>
<br>
All slides that have been presented in our NMLRG #3 meeting, June 27th, 201=
6, Athens, Greece, with EUCNC2016, have been uploaded. They can be accessed=
 through below link<br>
<br>
<a href=3D"https://www.ietf.org/proceedings/interim/2016/06/27/nmlrg/procee=
dings.html" rel=3D"noreferrer" target=3D"_blank">https://www.ietf.org/proce=
edings/interim/2016/06/27/nmlrg/proceedings.html</a><br>
<br>
Best regards,<br>
<br>
Sheng<br>
_______________________________________________<br>
nmlrg mailing list<br>
<a href=3D"mailto:nmlrg@irtf.org" target=3D"_blank">nmlrg@irtf.org</a><br>
<a href=3D"https://www.irtf.org/mailman/listinfo/nmlrg" rel=3D"noreferrer" =
target=3D"_blank">https://www.irtf.org/mailman/listinfo/nmlrg</a><br>
_______________________________________________<br>
nmlrg mailing list<br>
<a href=3D"mailto:nmlrg@irtf.org" target=3D"_blank">nmlrg@irtf.org</a><br>
<a href=3D"https://www.irtf.org/mailman/listinfo/nmlrg" rel=3D"noreferrer" =
target=3D"_blank">https://www.irtf.org/mailman/listinfo/nmlrg</a><br>
</blockquote>
</div>
<br>
</div>
</div>
</div>
</div>
</div>
</div>
</div>
</blockquote>
<div dir=3D"ltr" class=3D"gmail_extra"><br>
</div>
</div></div></div>
</div>
</div>
</div>

</blockquote></div><br></div></div>

--001a1140958411a5e30536850d34--


From voula.vassaki@incelligent.net  Thu Jun 30 12:53:54 2016
Return-Path: <voula.vassaki@incelligent.net>
X-Original-To: nmlrg@ietfa.amsl.com
Delivered-To: nmlrg@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 48DD412D5FD for <nmlrg@ietfa.amsl.com>; Thu, 30 Jun 2016 12:53:54 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.6
X-Spam-Level: 
X-Spam-Status: No, score=-2.6 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, HTML_MESSAGE=0.001, RCVD_IN_DNSWL_LOW=-0.7, SPF_PASS=-0.001] autolearn=ham autolearn_force=no
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id Lvu7tz-3sidU for <nmlrg@ietfa.amsl.com>; Thu, 30 Jun 2016 12:53:50 -0700 (PDT)
Received: from spamexperts-node1.grserver.gr (spamexperts-node1.grserver.gr [185.4.133.23]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by ietfa.amsl.com (Postfix) with ESMTPS id A095812D0BB for <nmlrg@irtf.org>; Thu, 30 Jun 2016 12:53:49 -0700 (PDT)
Received: from linux277.grserver.gr ([193.92.97.90]) by spamexperts-node1.grserver.gr with esmtps (TLSv1.2:DHE-RSA-AES256-SHA:256) (Exim 4.85) (envelope-from <voula.vassaki@incelligent.net>) id 1bIi1q-0000Ff-1Z; Thu, 30 Jun 2016 22:53:44 +0300
Received: from [127.0.0.1] (188.4.179.14.dsl.dyn.forthnet.gr [188.4.179.14]) by linux277.grserver.gr (Postfix) with ESMTPSA id CECC8834AA9; Thu, 30 Jun 2016 22:53:32 +0300 (EEST)
To: Panagiotis Demestichas <pdemestichas@gmail.com>, dmm@1-4-5.net
References: <5D36713D8A4E7348A7E10DF7437A4B927CA893E9@NKGEML515-MBX.china.huawei.com> <CAHiKxWi_mru0CG1ysoSVvN+beadHnmat2hzo5qeAQFudeYv75w@mail.gmail.com> <5D36713D8A4E7348A7E10DF7437A4B927CA89ED9@NKGEML515-MBX.china.huawei.com> <CAHiKxWgqqZbdAUkRbpTisjcW-K6HPOmapvMYcXtLfv5Z5X8i=A@mail.gmail.com> <CAKo0eZOgVEoejBAyMz44iANPD=2eGKvfOS7QCp6U7S_YUTUdew@mail.gmail.com> <CAKo0eZOMJadh2eHVDRz=gYa8RQBu5FJ393oFMx0amHSkg0KaLw@mail.gmail.com>
From: Voula Vassaki <voula.vassaki@incelligent.net>
Message-ID: <465e619c-0c2b-839b-95ea-e0c9da83149b@incelligent.net>
Date: Thu, 30 Jun 2016 22:53:30 +0300
User-Agent: Mozilla/5.0 (Windows NT 10.0; WOW64; rv:45.0) Gecko/20100101 Thunderbird/45.1.1
MIME-Version: 1.0
In-Reply-To: <CAKo0eZOMJadh2eHVDRz=gYa8RQBu5FJ393oFMx0amHSkg0KaLw@mail.gmail.com>
Content-Type: multipart/alternative; boundary="------------412F87F5A3245AE6AFB7E005"
X-Antivirus: avast! (VPS 160630-0, 06/30/2016), Outbound message
X-Antivirus-Status: Clean
X-PPP-Message-ID: <20160630195333.856345.90554@linux277.grserver.gr>
X-PPP-Vhost: incelligent.net
X-Filter-ID: s0sct1PQhAABKnZB5plbIa4A7F+Q2Ac3dlnk0wPzzfbTEr1WWGDk8VGWKBqZ77YzHmCjQuEGzBDV l11QyOfhrcG5Sp1rHiIKXoEXeEp49wuS6pzq/cnuvdbJuelATBvqUYKsju5yTprrB2odwarsb5eX BbVd7rmoEjvqFJrqqzqsiYhvWJqnYoAjX/y0bYhOJpODF/NPTtNCYu06b7IllqezLVvYs+1xhnv3 R0ZWbqB8kuky8OQbbtKFkKeqXJ6NmLXwaiDcD0KDUOdvWI6k4JM6u1vJhkjFkLkCGx1yFmPAsaoB /vOun7iN51GQztHKMh7Xl7hvRVG2lVyEWFbXXz3dKxLhoxcmaInYbR5vlqH6jL9AVZD0tI1Zo/Om dOqc0CpbsrErpKVHm2JCPPbSx9XDY1+sGIgb4JH2hOg5LFzyNtk2KJJ2teZf2+YHSfea790Jc7rx 0wAC/UDGwuwoez1jX4Hk0CX5SIpbwK5S75yXWxYGiuOTtpK5l5lufw0YRniheqlJYoiNeNBuh1ul bcb4nkq0y+sIFggiGPQK3WYm4uVZsXSCrHUkWMglPESVb4sI8sGKjhP8nv3TdoAzngnM7d1ITqcW y240lfsRXMe3uVo6NCLVB06ZobQmR6DBe6UxR5eG/Mk1+Z8NOGyD/KoCzutwPDdQfRK/w567yNrg rm3o9PQvoYky551B3D77ym2w78w4Hde2/v7eQ+VhTPgVB1xo6dCfBaU/iegBU8ajGqCefdc2LOvm NomA+msf+hnsOIx9tutuG16g49JxOgk4OCmwYa5v2FmmrwVripj2OKHH5lr9xXvSM4nM3avg
X-Report-Abuse-To: spam@spamexperts-node1.grserver.gr
X-Originating-IP: 193.92.97.90
X-SpamExperts-Domain: outgoing.grserver.gr
X-SpamExperts-Username: 193.92.97.90
Authentication-Results: grserver.gr; auth=pass smtp.auth=193.92.97.90@outgoing.grserver.gr
X-SpamExperts-Outgoing-Class: ham
X-SpamExperts-Outgoing-Evidence: SB/global_tokens (0.00623091384389)
X-Recommended-Action: accept
Archived-At: <https://mailarchive.ietf.org/arch/msg/nmlrg/oOUxUTri3PKg5ST64EEmCuhFG9Q>
X-Mailman-Approved-At: Thu, 30 Jun 2016 23:50:27 -0700
Cc: nmlrg@irtf.org, jiangsheng@huawei.com
Subject: Re: [Nmlrg] Fwd: links to each slide of presentations//RE: slides of NMLRG #3, June 27th, Athens
X-BeenThere: nmlrg@irtf.org
X-Mailman-Version: 2.1.17
Precedence: list
List-Id: Network Machine Learning Research Group <nmlrg.irtf.org>
List-Unsubscribe: <https://www.irtf.org/mailman/options/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=unsubscribe>
List-Archive: <https://mailarchive.ietf.org/arch/browse/nmlrg/>
List-Post: <mailto:nmlrg@irtf.org>
List-Help: <mailto:nmlrg-request@irtf.org?subject=help>
List-Subscribe: <https://www.irtf.org/mailman/listinfo/nmlrg>, <mailto:nmlrg-request@irtf.org?subject=subscribe>
X-List-Received-Date: Thu, 30 Jun 2016 19:55:17 -0000

This is a multi-part message in MIME format.
--------------412F87F5A3245AE6AFB7E005
Content-Type: text/plain; charset=utf-8; format=flowed
Content-Transfer-Encoding: 7bit


Dear  David,

as Prof. Demestichas explained the main reason of the classification is 
to be able to learn the characteristics of the traffic patterns in order 
to be able to allocate in an efficient manner the network resources. In 
case you need any further explanation regarding the presentation, do not 
hesitate to contact me.

Best regards,
Voula



>
> ---------- Forwarded message ----------
> From: *Panagiotis Demestichas* <pdemestichas@gmail.com 
> <mailto:pdemestichas@gmail.com>>
> Date: Thu, Jun 30, 2016 at 9:28 PM
> Subject: Re: [Nmlrg] links to each slide of presentations//RE: slides 
> of NMLRG #3, June 27th, Athens
> To: David Meyer <dmm@1-4-5.net <mailto:dmm@1-4-5.net>>
> Cc: Sheng Jiang <jiangsheng@huawei.com 
> <mailto:jiangsheng@huawei.com>>, "kt@incelligent.net 
> <mailto:kt@incelligent.net>" <kt@incelligent.net 
> <mailto:kt@incelligent.net>>, "nmlrg@irtf.org <mailto:nmlrg@irtf.org>" 
> <nmlrg@irtf.org <mailto:nmlrg@irtf.org>>
>
>
> Dear all
>
> just saw the stream of emails.
>
> The goal of the work is to learn traffic patterns, and other 
> associated aspects, e.g., in time and space. The goal is not to 
> instantaneously identify the class of the stream; such streams are 
> used for fine tuning the learning.
>
> Voula Vassaki (did the presentation) will complement.
>
> Best regards
> Panagiotis
>
>
>
> On Thu, Jun 30, 2016 at 8:58 PM, David Meyer <dmm@1-4-5.net 
> <mailto:dmm@1-4-5.net>> wrote:
>
>     Thanks Sheng, I meant it for the authors, but didn't see email
>     addresses in the deck...
>
>     --dmm
>
>
>     On Thu, Jun 30, 2016 at 10:44 AM, Sheng Jiang
>     <jiangsheng@huawei.com <mailto:jiangsheng@huawei.com>> wrote:
>
>         Hi, David,
>
>         I am not sure your question is for me or for the speaker.
>         Copied this email to the speaker, Panagiotis. If the question
>         is on me, I don't really understand the point of your
>         question. Could you please provide more context or explanation?
>
>         Thanks,
>
>         Sheng
>
>         ------------------------------------------------------------------------
>         *From:* David Meyer [dmm@1-4-5.net <mailto:dmm@1-4-5.net>]
>         *Sent:* 30 June 2016 2:09
>         *To:* Sheng Jiang
>         *Cc:* nmlrg@irtf.org <mailto:nmlrg@irtf.org>
>         *Subject:* Re: [Nmlrg] links to each slide of
>         presentations//RE: slides of NMLRG #3, June 27th, Athens
>
>         This question is on this use case:
>
>         Mobile network state characterization and prediction
>         https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-4.pdf
>
>         Isn't it the case that you will know if something is
>         registered to the MBB, MMC, or MCC slice at device
>         authentication time? If not what is the scenario in which you
>         don't know?
>
>         Thanks,
>
>         --dmm
>
>
>
>         On Tue, Jun 28, 2016 at 7:03 AM, Sheng Jiang
>         <jiangsheng@huawei.com <mailto:jiangsheng@huawei.com>> wrote:
>
>             Oops... The proceeding page for interim meeting seems not
>             be as intelligent as the proceeding pages of IETF
>             meetings. Our proceeding page does not autonomically show
>             the slides. I have sent an email to IETF secretary to ask
>             them to fix it. Meanwhile, in this email, here is the
>             links for each presentations:
>
>             Slide Filename Edit Replace Delete
>             Chair Slides
>             https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-0.pdf
>
>             Introduction to Network Machine Learning & NMLRG
>             https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-1.pdf
>
>             Data Collection and Analysis At High Security Lab
>             https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-2.pdf
>
>             Use Cases of Applying Machine Learning Mechanism with
>             Network Traffic
>             https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-3.pdf
>
>             Mobile network state characterization and prediction
>             https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-4.pdf
>
>             Learning how to route
>             https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-5.pdf
>
>             Regards,
>
>             Sheng
>             ________________________________________
>             From: nmlrg [nmlrg-bounces@irtf.org
>             <mailto:nmlrg-bounces@irtf.org>] on behalf of Sheng Jiang
>             [jiangsheng@huawei.com <mailto:jiangsheng@huawei.com>]
>             Sent: 28 June 2016 21:03
>             To: nmlrg@irtf.org <mailto:nmlrg@irtf.org>
>             Subject: [Nmlrg] slides of NMLRG #3, June 27th, Athens
>
>             Hi, nmlrg,
>
>             All slides that have been presented in our NMLRG #3
>             meeting, June 27th, 2016, Athens, Greece, with EUCNC2016,
>             have been uploaded. They can be accessed through below link
>
>             https://www.ietf.org/proceedings/interim/2016/06/27/nmlrg/proceedings.html
>
>             Best regards,
>
>             Sheng
>             _______________________________________________
>             nmlrg mailing list
>             nmlrg@irtf.org <mailto:nmlrg@irtf.org>
>             https://www.irtf.org/mailman/listinfo/nmlrg
>             _______________________________________________
>             nmlrg mailing list
>             nmlrg@irtf.org <mailto:nmlrg@irtf.org>
>             https://www.irtf.org/mailman/listinfo/nmlrg
>
>
>
>
>



---
This email has been checked for viruses by Avast antivirus software.
https://www.avast.com/antivirus

--------------412F87F5A3245AE6AFB7E005
Content-Type: text/html; charset=utf-8
Content-Transfer-Encoding: 8bit

<html>
  <head>
    <meta content="text/html; charset=utf-8" http-equiv="Content-Type">
  </head>
  <body bgcolor="#FFFFFF" text="#000000">
    <br>
    Dear  David, <br>
    <br>
    as Prof. Demestichas explained the main reason of the classification
    is to be able to learn the characteristics of the traffic patterns
    in order to be able to allocate in an efficient manner the network
    resources. In case you need any further explanation regarding the
    presentation, do not hesitate to contact me.<br>
    <br>
    Best regards, <br>
    Voula<br>
    <br>
    <br>
    <br>
    <blockquote
cite="mid:CAKo0eZOMJadh2eHVDRz=gYa8RQBu5FJ393oFMx0amHSkg0KaLw@mail.gmail.com"
      type="cite">
      <div dir="ltr">
        <div><br>
          <div class="gmail_quote">---------- Forwarded message
            ----------<br>
            From: <b class="gmail_sendername">Panagiotis Demestichas</b>
            <span dir="ltr">&lt;<a moz-do-not-send="true"
                href="mailto:pdemestichas@gmail.com">pdemestichas@gmail.com</a>&gt;</span><br>
            Date: Thu, Jun 30, 2016 at 9:28 PM<br>
            Subject: Re: [Nmlrg] links to each slide of
            presentations//RE: slides of NMLRG #3, June 27th, Athens<br>
            To: David Meyer &lt;<a moz-do-not-send="true"
              href="mailto:dmm@1-4-5.net">dmm@1-4-5.net</a>&gt;<br>
            Cc: Sheng Jiang &lt;<a moz-do-not-send="true"
              href="mailto:jiangsheng@huawei.com">jiangsheng@huawei.com</a>&gt;,
            "<a moz-do-not-send="true" href="mailto:kt@incelligent.net">kt@incelligent.net</a>"
            &lt;<a moz-do-not-send="true"
              href="mailto:kt@incelligent.net">kt@incelligent.net</a>&gt;,
            "<a moz-do-not-send="true" href="mailto:nmlrg@irtf.org">nmlrg@irtf.org</a>"
            &lt;<a moz-do-not-send="true" href="mailto:nmlrg@irtf.org">nmlrg@irtf.org</a>&gt;<br>
            <br>
            <br>
            <div dir="ltr">Dear all<br>
              <div><br>
              </div>
              <div>just saw the stream of emails.</div>
              <div><br>
              </div>
              <div>The goal of the work is to learn traffic patterns,
                and other associated aspects, e.g., in time and space.
                The goal is not to instantaneously identify the class of
                the stream; such streams are used for fine tuning the
                learning.</div>
              <div><br>
              </div>
              <div>
                <div>Voula Vassaki (did the presentation) will
                  complement.</div>
              </div>
              <div><br>
              </div>
              <div>Best regards</div>
              <div>Panagiotis</div>
              <div><br>
              </div>
              <div><br>
              </div>
            </div>
            <div class="gmail_extra"><br>
              <div class="gmail_quote">On Thu, Jun 30, 2016 at 8:58 PM,
                David Meyer <span dir="ltr">&lt;<a
                    moz-do-not-send="true" href="mailto:dmm@1-4-5.net"
                    target="_blank"><a class="moz-txt-link-abbreviated" href="mailto:dmm@1-4-5.net">dmm@1-4-5.net</a></a>&gt;</span> wrote:<br>
                <blockquote class="gmail_quote" style="margin:0 0 0
                  .8ex;border-left:1px #ccc solid;padding-left:1ex">
                  <div dir="ltr">Thanks Sheng, I meant it for the
                    authors, but didn't see email addresses in the
                    deck...
                    <div><br>
                    </div>
                    <div>--dmm</div>
                    <div><br>
                      <div class="gmail_extra"><br>
                        <div class="gmail_quote">On Thu, Jun 30, 2016 at
                          10:44 AM, Sheng Jiang <span dir="ltr">&lt;<a
                              moz-do-not-send="true"
                              href="mailto:jiangsheng@huawei.com"
                              target="_blank"><a class="moz-txt-link-abbreviated" href="mailto:jiangsheng@huawei.com">jiangsheng@huawei.com</a></a>&gt;</span>
                          wrote:<br>
                          <blockquote class="gmail_quote"
                            style="margin:0 0 0 .8ex;border-left:1px
                            #ccc solid;padding-left:1ex">
                            <div>
                              <div
                                style="FONT-FAMILY:Tahoma;DIRECTION:ltr;COLOR:#000000;FONT-SIZE:10pt">
                                <p>Hi, David,</p>
                                <p> </p>
                                <p>I am not sure your question is for me
                                  or for the speaker. Copied this email
                                  to the speaker, Panagiotis. If the
                                  question is on me, I don't really
                                  understand the point of your question.
                                  Could you please provide more context
                                  or explanation?</p>
                                <p> </p>
                                <p>Thanks,</p>
                                <p> </p>
                                <p>Sheng</p>
                                <p> </p>
                                <div style="FONT-FAMILY:Times New
                                  Roman;COLOR:#000000;FONT-SIZE:16px">
                                  <hr>
                                  <div style="DIRECTION:ltr"><font
                                      face="Tahoma" color="#000000"
                                      size="2"><b>From:</b> David Meyer
                                      [<a moz-do-not-send="true"
                                        href="mailto:dmm@1-4-5.net"
                                        target="_blank">dmm@1-4-5.net</a>]<br>
                                      <b>Sent:</b> 30 June 2016 2:09<span><br>
                                        <b>To:</b> Sheng Jiang<br>
                                        <b>Cc:</b> <a
                                          moz-do-not-send="true"
                                          href="mailto:nmlrg@irtf.org"
                                          target="_blank"><a class="moz-txt-link-abbreviated" href="mailto:nmlrg@irtf.org">nmlrg@irtf.org</a></a><br>
                                      </span><b>Subject:</b> Re: [Nmlrg]
                                      links to each slide of
                                      presentations//RE: slides of NMLRG
                                      #3, June 27th, Athens<br>
                                    </font><br>
                                  </div>
                                  <div>
                                    <div>
                                      <div>
                                        <div dir="ltr">This question is
                                          on this use case:
                                          <div><br>
                                            <div>
                                              <div>Mobile network state
                                                characterization and
                                                prediction</div>
                                              <div><a
                                                  moz-do-not-send="true"
href="https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-4.pdf"
                                                  target="_blank"><a class="moz-txt-link-freetext" href="https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-4.pdf">https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-4.pdf</a></a></div>
                                            </div>
                                            <div><br>
                                            </div>
                                            <div>Isn't it the case that
                                              you will know if something
                                              is registered to the MBB,
                                              MMC, or MCC slice at
                                              device authentication
                                              time? If not what is the
                                              scenario in which you
                                              don't know?</div>
                                            <div><br>
                                            </div>
                                            <div>Thanks,</div>
                                            <div><br>
                                            </div>
                                            <div>--dmm</div>
                                            <div><br>
                                            </div>
                                            <div><br>
                                              <div class="gmail_extra"><br>
                                                <div class="gmail_quote">On
                                                  Tue, Jun 28, 2016 at
                                                  7:03 AM, Sheng Jiang <span
                                                    dir="ltr">
                                                    &lt;<a
                                                      moz-do-not-send="true"
href="mailto:jiangsheng@huawei.com" target="_blank"><a class="moz-txt-link-abbreviated" href="mailto:jiangsheng@huawei.com">jiangsheng@huawei.com</a></a>&gt;</span>
                                                  wrote:<br>
                                                  <blockquote
                                                    style="BORDER-LEFT:rgb(204,204,204)
                                                    1px solid;MARGIN:0px
                                                    0px 0px
                                                    0.8ex;PADDING-LEFT:1ex"
                                                    class="gmail_quote">
                                                    Oops... The
                                                    proceeding page for
                                                    interim meeting
                                                    seems not be as
                                                    intelligent as the
                                                    proceeding pages of
                                                    IETF meetings. Our
                                                    proceeding page does
                                                    not autonomically
                                                    show the slides. I
                                                    have sent an email
                                                    to IETF secretary to
                                                    ask them to fix it.
                                                    Meanwhile, in this
                                                    email, here is the
                                                    links for each
                                                    presentations:<br>
                                                    <br>
                                                    Slide Filename Edit
                                                    Replace Delete<br>
                                                    Chair Slides<br>
                                                    <a
                                                      moz-do-not-send="true"
href="https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-0.pdf"
                                                      rel="noreferrer"
                                                      target="_blank"><a class="moz-txt-link-freetext" href="https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-0.pdf">https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-0.pdf</a></a><br>
                                                    <br>
                                                    Introduction to
                                                    Network Machine
                                                    Learning &amp; NMLRG<br>
                                                    <a
                                                      moz-do-not-send="true"
href="https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-1.pdf"
                                                      rel="noreferrer"
                                                      target="_blank"><a class="moz-txt-link-freetext" href="https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-1.pdf">https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-1.pdf</a></a><br>
                                                    <br>
                                                    Data Collection and
                                                    Analysis At High
                                                    Security Lab<br>
                                                    <a
                                                      moz-do-not-send="true"
href="https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-2.pdf"
                                                      rel="noreferrer"
                                                      target="_blank"><a class="moz-txt-link-freetext" href="https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-2.pdf">https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-2.pdf</a></a><br>
                                                    <br>
                                                    Use Cases of
                                                    Applying Machine
                                                    Learning Mechanism
                                                    with Network Traffic<br>
                                                    <a
                                                      moz-do-not-send="true"
href="https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-3.pdf"
                                                      rel="noreferrer"
                                                      target="_blank"><a class="moz-txt-link-freetext" href="https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-3.pdf">https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-3.pdf</a></a><br>
                                                    <br>
                                                    Mobile network state
                                                    characterization and
                                                    prediction<br>
                                                    <a
                                                      moz-do-not-send="true"
href="https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-4.pdf"
                                                      rel="noreferrer"
                                                      target="_blank"><a class="moz-txt-link-freetext" href="https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-4.pdf">https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-4.pdf</a></a><br>
                                                    <br>
                                                    Learning how to
                                                    route<br>
                                                    <a
                                                      moz-do-not-send="true"
href="https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-5.pdf"
                                                      rel="noreferrer"
                                                      target="_blank"><a class="moz-txt-link-freetext" href="https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-5.pdf">https://www.ietf.org/proceedings/interim-2016-nmlrg-01/slides/slides-interim-2016-nmlrg-01-5.pdf</a></a><br>
                                                    <br>
                                                    Regards,<br>
                                                    <br>
                                                    Sheng<br>
________________________________________<br>
                                                    From: nmlrg [<a
                                                      moz-do-not-send="true"
href="mailto:nmlrg-bounces@irtf.org" target="_blank"><a class="moz-txt-link-abbreviated" href="mailto:nmlrg-bounces@irtf.org">nmlrg-bounces@irtf.org</a></a>]
                                                    on behalf of Sheng
                                                    Jiang [<a
                                                      moz-do-not-send="true"
href="mailto:jiangsheng@huawei.com" target="_blank"><a class="moz-txt-link-abbreviated" href="mailto:jiangsheng@huawei.com">jiangsheng@huawei.com</a></a>]<br>
                                                    Sent: 28 June 2016
                                                    21:03<br>
                                                    To: <a
                                                      moz-do-not-send="true"
href="mailto:nmlrg@irtf.org" target="_blank"><a class="moz-txt-link-abbreviated" href="mailto:nmlrg@irtf.org">nmlrg@irtf.org</a></a><br>
                                                    Subject: [Nmlrg]
                                                    slides of NMLRG #3,
                                                    June 27th, Athens<br>
                                                    <br>
                                                    Hi, nmlrg,<br>
                                                    <br>
                                                    All slides that have
                                                    been presented in
                                                    our NMLRG #3
                                                    meeting, June 27th,
                                                    2016, Athens,
                                                    Greece, with
                                                    EUCNC2016, have been
                                                    uploaded. They can
                                                    be accessed through
                                                    below link<br>
                                                    <br>
                                                    <a
                                                      moz-do-not-send="true"
href="https://www.ietf.org/proceedings/interim/2016/06/27/nmlrg/proceedings.html"
                                                      rel="noreferrer"
                                                      target="_blank"><a class="moz-txt-link-freetext" href="https://www.ietf.org/proceedings/interim/2016/06/27/nmlrg/proceedings.html">https://www.ietf.org/proceedings/interim/2016/06/27/nmlrg/proceedings.html</a></a><br>
                                                    <br>
                                                    Best regards,<br>
                                                    <br>
                                                    Sheng<br>
_______________________________________________<br>
                                                    nmlrg mailing list<br>
                                                    <a
                                                      moz-do-not-send="true"
href="mailto:nmlrg@irtf.org" target="_blank"><a class="moz-txt-link-abbreviated" href="mailto:nmlrg@irtf.org">nmlrg@irtf.org</a></a><br>
                                                    <a
                                                      moz-do-not-send="true"
href="https://www.irtf.org/mailman/listinfo/nmlrg" rel="noreferrer"
                                                      target="_blank"><a class="moz-txt-link-freetext" href="https://www.irtf.org/mailman/listinfo/nmlrg">https://www.irtf.org/mailman/listinfo/nmlrg</a></a><br>
_______________________________________________<br>
                                                    nmlrg mailing list<br>
                                                    <a
                                                      moz-do-not-send="true"
href="mailto:nmlrg@irtf.org" target="_blank"><a class="moz-txt-link-abbreviated" href="mailto:nmlrg@irtf.org">nmlrg@irtf.org</a></a><br>
                                                    <a
                                                      moz-do-not-send="true"
href="https://www.irtf.org/mailman/listinfo/nmlrg" rel="noreferrer"
                                                      target="_blank"><a class="moz-txt-link-freetext" href="https://www.irtf.org/mailman/listinfo/nmlrg">https://www.irtf.org/mailman/listinfo/nmlrg</a></a><br>
                                                  </blockquote>
                                                </div>
                                                <br>
                                              </div>
                                            </div>
                                          </div>
                                        </div>
                                      </div>
                                    </div>
                                  </div>
                                </div>
                              </div>
                            </div>
                          </blockquote>
                        </div>
                        <br>
                      </div>
                    </div>
                  </div>
                </blockquote>
              </div>
              <br>
            </div>
          </div>
          <br>
        </div>
      </div>
    </blockquote>
    <br>
  <br />
<table style="border-top: 1px solid #D3D4DE;">
	<tr>
        <td style="width: 55px; padding-top: 13px;"><a href="https://www.avast.com/sig-email?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=emailclient&utm_term=icon" target="_blank"><img src="https://ipmcdn.avast.com/images/2016/icons/icon-envelope-tick-round-orange_184x116-v1.png" width="46" height="29" style="width: 46px; height: 29px;" /></a></td>
		<td style="width: 470px; padding-top: 12px; color: #41424e; font-size: 13px; font-family: Arial, Helvetica, sans-serif; line-height: 18px;">Virus-free. <a href="https://www.avast.com/sig-email?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=emailclient&utm_term=link" target="_blank" style="color: #4453ea;">www.avast.com</a>
		</td>
	</tr>
</table></body>
</html>

--------------412F87F5A3245AE6AFB7E005--

