NMRG 62nd meeting
22 June 2021 14:00 - 15:00 CEST


RG Chairs:
* Laurent Ciavaglia
* Jéröme François

RG Secretaries
* Jéferson Campos Nobre
* Pedro Martinez-Julia

Useful links:
Materials: https://datatracker.ietf.org/meeting/interim-2021-nmrg-03/session/nmrg
Webex: https://ietf.webex.com/ietf/j.php?MTID=mf4f7414b4e2928cf35fb18b957a29c50 
Notes: https://codimd.ietf.org/notes-nmrg62# 
Video recording: https://www.youtube.com/user/ietf/playlists


Participants:
- Jérôme François
- Alex Clemm
- Hyun-Kyo Kim
- Chiara Contoli
- Jonatas Marques
- Stefan Schneider
- Marinos Charalambides
- Börje Ohlman
- José Suárez-Varela
- Ramin Sadre
- Lisandro Granville

Objectives:
- Freeze the challenge list
- Set the expected content type and levels of details for challenges → template
- Define one lead contributor per challenge
- Set milestones for progressing on the document

Discussion
- Alex:  Freeze the challenge for a fist iteration but we can add in a second round
- Alex (on IBN challenges): challenge #9 is about how AI can facilitate the planning of actions. Can rephrase to make it clear
- Stefan: there are different type of outputs for AI, in challenge #9
- Marinos: is dealing with with different choices of actions to take part of #9
- Alex: ok
- Jérôme: challenge #9 is fine, we should avoid to make specific to IBN but more general about planning of actions. We will need more feedback regarding challenge #8
- Marinos (On merging data availability and challenge on scalable real-time monitoring) ok
- Alex: having data is an enabler for AI but not an AI challenge?
- Jérôme: agrre but here we try to identify challenges to address when we want to use AI for NM, so I think it is good to have it as well
- Stefan: I agree, data is an enabler, but I think it's a very important one and one that is indeed challenging in the context of NM. For image classification, NLP, etc there are standard datasets or environments that drive progress; in NM, we don't have that yet.
- Alex: Agree. So then it would be good to frame that as an AI challenge: how to create the equivalent in NM (in terms of standard datsets or environments that can help drive AI progress in NM
- Alex: maybe we could have different sections in the document (What AI can solve in NM? What are the missing enabblers?)
- Jérôme: yes challenges are at different levels. There was a draft classification in the doc. I propose to kee the structure of challenge list flat and then categorize them when they are fully describe.
- Lisandro (regarding mislestones): It looks to be that's a doable timeline
- José Suárez-Varela: It looks good to me as wel