Re: [Idnet] IETF99 for applying AI/ML into network management: Follow-up

[Pedro Martinez-Julia <[e-mail]>]

On Thu, Jul 20, 2017 at 07:05:17PM +0200, Jérôme François wrote:
> Dear all,

Dear Jerome,

Thank you very much for your comments. Please find below some hints from
my side.

> Thanks Sheng for having organized the meeting. As we are running out of
> time, here are some comments I want to share:
> - in the figure with the loop, data to be used could be from other
> sources of data out from the network to enrich the analysis

This is delicate since, strictly speaking, such data cannot form part of
the "control loop" because it is not a variable to control. However, in
a practical view it will be connected to the input of the loop (and I am
currently working on solutions that consider such information in such
point), but it must be clear that it will not form part of the closure
of the loop.

> - regarding use cases, it was unclear for me what is your or the group
> plan. I think many of us express interests for other use cases but I'm
> not sure how to contribute / select use cases (and if we have to ? ->
> see next item)

To my experience, the best option is to find some common requirements to
all use cases we are taking into account and then try to find a smaller
use case that reflects only those requirements. We can then "vote" which
aspects must be considered from the beginning and which can be left for
future use cases.

> - the immediate plan is to organize a regular workshop (probably within
> NMRG). I fully support the idea as this helps to federate the community
> first to ambition then coordinated actions (regarding the issues with
> datasets for example). But recall that that will be workshop within a RG
> group and it is probably too early to identify what should be
> standardized  (or look at a WG in that case). Looking at the experience
> with flow measurement workshop (within NMRG as well), it should be more
> open for presentations and discussions and if a potential standard is
> identified, this can be targeted afterwards.

That is nice, but has some risks. We must first set a strong topic or we
can find a too high diversion from the issues that IDNET must resolve.

I agree with you in that it is good to gather proposals to resolve the
dataset issues and it can be the main (I propose that only) topic of the
first workshop on IDNET.

Those are my 2 cents. Thanks for reading.

> Best regards,
> jerome

Regards,
Pedro

-- 
Pedro Martinez-Julia
Network Science and Convergence Device Technology Laboratory
Network System Research Institute
National Institute of Information and Communications Technology (NICT)
4-2-1, Nukui-Kitamachi, Koganei, Tokyo 184-8795, Japan
Email: [e-mail]
---------------------------------------------------------
*** Entia non sunt multiplicanda praeter necessitatem ***