Re: [nmrg] [NMRG] research challenges in AI for NM - Next meeting

["Ciavaglia, Laurent (Nokia - FR/Paris-Saclay)" <[e-mail]>]

Hi Jérôme,

Looking forward to the meeting next week to progress on the document.

Regarding your question on how to rightly define the challenges:
	- An important thing to clarify early is to avoid confusion between a challenge and a use case.
	A rough distinction would be that a challenge exists at the NM domain level and applies across multiple/all use cases. A use case may illustrate several challenges, or facets of challenges in a particular context.
	I think that we could utilize use cases to fuel our analysis and reflections but also that we should strive to extract commonalities, patterns among the use cases. Such commonalities could hint to more systemic challenges, or help in their identification, understanding and definition.
	I agree with the approach to analyze the characteristics of NM problems or activities (based on a "taxonomy", e.g. FCAPS) and analyze the characteristics of AI techniques; however we should be careful not to prescribe or imply which AI technique(s) should be used to address which NM problem(s).
	- Another aspect that could help to properly describe (the scope of the) challenges is to define what/how we aim to describe them, e.g. a description, motivations for, considerations of the challenge followed by research questions or directions. This would be in line with the existing guidelines for challenge description.
	- Sources to identify and define the right set of challenges could be diverse: SotA review, technical documentation (e.g. from standards, open source...), but also I think we should not neglect our own domain knowledge and capacity of analysis (e.g. by asking ourselves questions such as: "Is the list of challenges comprehensive?" "Are there gaps or additional areas to consider?" "Are we covering the most important aspects of AI-based NM?", etc.)

Best regards, 
Laurent

> -----Original Message-----
> From: Jérôme François <[e-mail]>
> Sent: Tuesday, December 8, 2020 22:49
> To: [e-mail]
> Cc: [e-mail]
> Subject: [NMRG] research challenges in AI for NM - Next meeting
> 
> Dear all,
> 
> The next meeting regarding the AI document will be Dec 15 at 14:00 (UTC+1)
> 
> Tentative agenda:
> - review comments received regarding scope
> - discuss the method on how to identify the righ challenge sets (and so
> verify the
> oens we already described): I was suggesting to list properties of NM
> problems and
> AI technique properties) -> if anyone has good references on theses
> aspects,
> please share it.
> 
> 
> Any other suggestion ?
> 
> A datatracker entry will be created for this meeting. Webex details below.
> 
> Best regards
> jerome
> 
> 
> 
> https://ietf.webex.com/ietf/j.php?MTID=m7d9057e767719e6fcd1b7d1f05975e0f
> Tuesday, 15 Dec, 2020 14:00 | 1 hour | (UTC+01:00) Brussels, Copenhagen,
> Madrid, Paris
> Meeting number: 178 859 5216
> Password: cEP6WaMNu33
> 432c93cde3704433833ae5836e94a06f
> 
> Join by video system
> Dial [e-mail]
> You can also dial 173.243.2.68 and enter your meeting number.
> 
> Join by phone
> 1-650-479-3208 Call-in toll number (US/Canada)
> Access code: 178 859 5216