Re: [Nmlrg] Closedown of the proposed NMLRG and NMLRG mail list

Very quick input ... won't be eloquent but I hope is reasonably intelligible:

- I agree with the value of expanding the scope to AI, writ large, possibly under a different name (but I don't shy away from the AI label myself).
- Following a path from Automation -> Autonomics -> Autonomy, where the path is bounded (i.e., Pedro's "borders") via "_instrumented_ governance" (by humans), is one way to ease fears and minimize risks.
- I agree with the need to include ontology ... mainly because that should enforce thinking up front about the kinds of reasoning operations that will facilitate trust in the borders and instrumented governance.

Avanti,
BobN

-----Original Message-----
From: nmlrg [mailto:nmlrg-bounces@irtf.org] On Behalf Of Pedro Martinez-Julia
Sent: Thursday, December 08, 2016 2:30 AM
To: Sheng Jiang <jiangsheng@huawei.com>
Cc: nmlrg@irtf.org; Albert Cabellos <albert.cabellos@gmail.com>
Subject: Re: [Nmlrg] Closedown of the proposed NMLRG and NMLRG mail list

On Thu, Dec 08, 2016 at 06:28:43AM +0000, Sheng Jiang wrote:
> Hi, Pedro,

Hi,

> I fully agree that AI is wider than ML although it is more difficult 
> to give a definition or border of AI technologies. We can certainly 
> extend our scope from ML to AI. On other side, our scope is actually 
> wider than network management. We also targeted autonomic network 
> controlling from the very beginning.

I was tempted to mention "network control" but, from my point of view, management is a broader term that can somehow include control. I agree that Autonomic Networking (AN) would be a direct beneficiary of AI, and therefore it gives more sense to the extended scope of the group.

Regarding the definition and border, from my point of view, they would be naturally derived from the specific application of AI techniques to network control and management. In fact, if we start defining the set of concepts and relations I mentioned (more specifically, an ontology), it would be clear where to establish the specific borders. However, almost any AI technique could be useful in this context.

On the other hand, and maybe a bit against my own proposal, we have to face the concerns (fears?) that many people have about AI mechanisms taking decisions autonomously. We can begin with the definition of some deployment plan (better name, please) that incrementally gives autonomy to the final system, together with a set of rules (standard?) that must be set to every AI-based management and control system. As a note, it can follow the steps of the autonomous cars that are already being left driving alone, also taking into account and learning from their flaws and problems.

Finally, if many group members agree on extending the scope to AI, it would be good to begin right now, with the creation of the new mailing list, and (of course) set a proper name to it :-D.

Let's be in contact.

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: pedro@nict.go.jp
---------------------------------------------------------
*** Entia non sunt multiplicanda praeter necessitatem ***

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