Re: [Nmlrg] One comment on the proposed charter...

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.

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