NPPS Leadership Team

[Brett Viren <bv AT bnl.gov>]

Hi Torre,

sPHENIX (Jin) + CSI (Yi Huang) have a paper on AI/ML compression from
last year (or perhaps it was 2022).  It may be useful for the proposal.

If compression error and bias are in scope (and I guess they really must
be) perhaps there are some ideas from the LS4GAN LDRD that could be
added to the proposal.  This would let LS4GAN point to "new funding"
that came from the LDRD spending and the added facets may help the new
proposal succeed.

At the least there is some applicable "experience" that could transfer.
But perhaps more interesting are some new and untested ideas.  As they
are untested, developing them as part of a new proposal would not be
"double dipping".

They involve attempting to tie apparent (eg pixel-wise) errors in
translation (compression) to known a'priori errors in the physics
parameters that make the input to simulated samples.  Of course, sim is
biased relative to real detector data but then with LS4GAN we may be
able to translate these physics-based errors to the real detector data
domain.  How to estimate the "error in the error" due to this
translation is still an unknown (to me) but perhaps that can be
overcome.

As is probably obvious, this is all in a very hand-wavy stage and it may
end up being pure vapor but if it seems useful we can discuss more.


-Brett.


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

> The one expression of interest re: this call that I've heard is from JLab 
> :-) I've said to Markus I think it's a
> good idea.
>   Torre
>
> ---------- Forwarded message ---------
> From: Markus Diefenthaler <mdiefent AT jlab.org>
> Date: Thu, Jan 25, 2024 at 8:00 PM
> Subject: Fwd: [EXTERNAL] Data Reduction for Science Funding Opportunity
> To: Torre Wenaus <wenaus AT bnl.gov>
>
> Please see below. Should we submit a proposal for streaming data processing 
> at ePIC and the related data reduction
> from time slices to reconstructed events)? 
>
>     Begin forwarded message:
>    
>     From: "DOE Office of Science" <updates AT info.science.doe.gov>
>     Subject: [EXTERNAL] Data Reduction for Science Funding Opportunity
>     Date: January 18, 2024 at 11:16:41 EST
>     To: mdiefent AT jlab.org
>     Reply-To: updates AT info.science.doe.gov
>
>      View as a webpage / Share                                              
>                                         
>                                                                             
>                                         
>                                            U.S. Department of Energy - 
> Office                                       
>
>                                            Funding Opportunity 
> Announcement:                                        
>                                                                             
>                                         
>                                      Data Reduction for Science Funding 
> Opportunity                                 
>                                                                             
>                                         
>                              Funding Opportunity Announcement (FOA) Number: 
> DE-FOA-0003266                          
>                                                                             
>                                         
>                                           Total Estimated Funding: $15 
> Million                                      
>                                                                             
>                                         
>      
> ┌───────────────────────────────────────────┬────────────────────────────────┐
>                                  
>      │Deadline for Letters of Intent (required): │ March 19, 2024 at 
> 11:59pm ET   │                                 
>      
> ├───────────────────────────────────────────┼────────────────────────────────┤
>                                  
>      │Deadline for Applications:                 │ May 7, 2024 at 11:59pm 
> ET      │                                 
>      
> └───────────────────────────────────────────┴────────────────────────────────┘
>                                  
>                                                                             
>                                         
>      The U.S. Department of Energy’s Office of Science, under the Advanced 
> Scientific Computing Research Program,   
>      is announcing $15 million in available funding to support the 
> advancement of data reduction for science. This  
>      research will explore potentially high-impact approaches to develop 
> and use data reduction techniques and      
>      algorithms to facilitate more efficient analysis and use of massive 
> data sets produced by observations,        
>      experiments, and simulation. These different types of sources are 
> producing data at rates beyond current       
>      capacity to store, analyze, stream, and archive it in raw form. As a 
> result, many research groups have begun   
>      reducing the size of their data sets via techniques such as 
> compression, reduced order models,                 
>      experiment-specific triggers, filtering, and feature extraction.       
>                                         
>                                                                             
>                                         
>      This research program seeks to continue to increase the level of 
> mathematical rigor in scientific data         
>      reduction to ensure that scientifically relevant constraints on 
> quantities of interest are satisfied, methods  
>      can be integrated into scientific workflows, and methods are 
> implemented in a manner that inspires trust that  
>      the desired information is preserved. Data is ubiquitous for every 
> scientific discipline, and it is            
>      foundational to the recent, current, and future advancements in 
> scientific machine learning and artificial     
>      intelligence. Machine learning is particularly ripe for data reduction 
> advances, as data reduction can improve 
>      the efficiency of learning, and machine learning techniques can be 
> used to reduce data.                        
>                                                                             
>                                         
>      Learn more about this funding opportunity announcement and eligibility 
> by visiting the website.                
>                                                                             
>                                         
>      View Funding Opportunity                                               
>                                         
>                                                                             
>                                         
>      U.S. Department of Energy - Office of Science                          
>                                         
>                                                                             
>                                         
>      You  Twi  RSS                                                          
>                                         
>                                                                             
>                                         
>                                    Manage Subscriptions  |  Unsubscribe All 
>  |  Help                                
>                                                                             
>                                         
>     
> ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
>     This email was sent to mdiefent AT jlab.org using GovDelivery 
> Communications Cloud on behalf of: US       GovDelive
>     Department of Energy Office of Science · 1000 Independence Ave., SW · 
> Washington, DC · 20585 · (202)            
>     586-5430                                                                
>                                         
>     *
>
> --
> -- Torre Wenaus, BNL NPPS Group, ATLAS Experiment
> -- BNL 510A 1-222 | 631-681-7892
> _______________________________________________
> Phys-npps-mgmt-l mailing list
> Phys-npps-mgmt-l AT lists.bnl.gov
> https://lists.bnl.gov/mailman/listinfo/phys-npps-mgmt-l

Attachment: signature.asc
Description: PGP signature