NPPS Leadership Team

["Lin, Meifeng" <mlin AT bnl.gov>]

Hi Torre,

 

I also think that having a “BNL AI/ML Strategy” slide would be nice, and you can mention how the BNL HEP AI/ML strategy would fit into the lab-wide strategy.

 

I second Michael’s comment on the unfunded LDRDs, especially if the slides will feed into future HEP funding calls. It is important to make the HEP office aware of our directions.  

 

I also made some comments directly on the slides.

 

Meifeng

 

-- 
Meifeng Lin, Ph.D.

Group Leader, High Performance Computing

Computational Science Initiative
Brookhaven National Laboratory

Email: mlin AT bnl.gov

https://www.bnl.gov/staff/mlin

 

 

 

From: Alexei Klimentov <alexei.klimentov AT cern.ch>
Date: Friday, September 24, 2021 at 3:03 AM
To: Michael Begel <mbegel AT gmail.com>
Cc: Viren, Brett <bviren AT bnl.gov>, Alexander, Francis <falexander AT bnl.gov>, Lin, Meifeng <mlin AT bnl.gov>, Torre Wenaus via Phys-npps-mgmt-l <phys-npps-mgmt-l AT lists.bnl.gov>, Aschenauer, Elke <elke AT bnl.gov>, Pleier, Marc-Andre <mpleier AT bnl.gov>
Subject: Re: [Phys-npps-mgmt-l] BNL HEP AI/ML vision draft for your comments

Hi Torre,

 

 I agree with Brett that ’BNL strategy’ slide/bullets will be useful, also you can add ML/AL tools for a mono-H(bb)-MET in ATLAS, it is in production (on slide 4)

 

Cheers, Alexei 

On Sep 23, 2021, at 7:48 PM, Michael Begel via Phys-npps-mgmt-l <phys-npps-mgmt-l AT lists.bnl.gov> wrote:

 

I agree with Brett. This doesn't come across as overarching "strategy." Especially BNL leadership being 5-10 years out. Perhaps this should be compressed: 1, 2-3, 4-5?

 

FYI: the Cognizent DAQ LDRD wasn't funded. I don't know if the others were funded or not. But I'd still leave this in the document as it illustrates our thinking.

 

Michael Begel

 

 

On Thu, Sep 23, 2021 at 9:55 AM Brett Viren <bv AT bnl.gov> wrote:

Hi Torre,

This is perhaps a bad idea given the reality of slide 3 but....

Is it beneficial to express some "BNL AI/ML strategy"?  Especially if it
sets us apart from the unwashed masses?

It is widely recognized (and seems to be a recent trending Internet
meme) that one should first exhaust "conventional" ways to extract
information from data before applying AI/ML.  Use accrued domain
knowledge to wisely guide application of AI/ML.

This is a strategy our group has adhered to strongly for the LArTPC work
while our other neutrino LArTPC AI/ML friends prefer to throw data at
the GPU and see what slides off.

I feel it is the right strategy, but it goes against funding trends
where real thinking is being discounted in favor of the artificial type.

So, again, maybe a bad suggestion....


Otherwise, I think the slides are good.

-Brett.

Torre Wenaus via Phys-npps-mgmt-l <phys-npps-mgmt-l AT lists.bnl.gov>
writes:

> Hi all,
>
> The next monthly discussion between DOE HEP and BNL HEP on Oct 1 will address AI/ML. I’ll give a short talk, along the
> lines of recent 5min ones at the Science Council and AI/ML working group kickoff, but a bit longer so some time for
> examples to flesh out the generic statements. The working group kickoff version of the slides is at 
>
>    https://docs.google.com/presentation/d/15ElWnNopunK86wijlWd78yNjWfJDw-ckX_EEoaTUuAA/edit#slide=id.geaae160bb5_0_0 
>
> (The DOE talk will for sure not have slide 3.)
>
> The basis for the DOE slides will be an update of a draft BNL HEP AI/ML vision 2-pager I wrote a few months ago at
> Dmitri’s request, the main difference in the update being the incorporation of examples.
>
> Here is the updated version, your comments/corrections (by the end of the week if you can!) would be much appreciated.
>
>   https://docs.google.com/document/d/11fnQxO4Az23ADQQBhIXv2B-fxOhzOLunj4WnXtz3ioE/edit?usp=sharing 
>
>   Torre
>
> --
> -- Torre Wenaus, BNL NPPS Group, ATLAS Experiment
> -- BNL 510A 1-222 | 631-681-7892 |  wenaus AT gmail.com | npps.bnl.gov | wenaus.com
> -- NPPS Mattermost room: https://chat.sdcc.bnl.gov/npps/channels/town-square
> _______________________________________________
> Phys-npps-mgmt-l mailing list
> Phys-npps-mgmt-l AT lists.bnl.gov
> https://lists.bnl.gov/mailman/listinfo/phys-npps-mgmt-l

_______________________________________________
Phys-npps-mgmt-l mailing list
Phys-npps-mgmt-l AT lists.bnl.gov
https://lists.bnl.gov/mailman/listinfo/phys-npps-mgmt-l