1 of 8

Workshop overview

The organizers

2 of 8

History

Feb 2019

Sept 2019

April 2019, MIT workshop

https://indico.cern.ch/event/714134/

3 of 8

Why we’re here...

Cutting edge science requires:

Faster, more precise, bigger, more granular,...

Real-time, accelerated machine learning can greatly accelerate time to science, allowing us to:

  • test hypotheses significantly faster
  • enhance and automate performance of detectors/accelerators
  • save and maximize potentially lost data

​

4 of 8

A Multi-

Disciplinary

Workshop

Scientific domain

applications

Machine learning techniques

Compute elements & systems

YOU ARE HERE!

Domain-�inspired ML

Science�data�pipelines

ML-specific�systems

5 of 8

A Multi-

Disciplinary

Workshop

Scientific domain

applications

Machine learning techniques

Compute elements & systems

YOU ARE HERE!

Domain-�inspired ML

Science�data�pipelines

ML-specific�systems

6 of 8

What we hope you get out of this workshop

Hear from experts about

  • Status and goals in Fast ML for a number scientific domains
  • Advances in computing technology & hardware
  • Efficient machine learning in resource constrained environments
  • Exciting new work on applications of Fast ML

​

Supplement workshop with forward-looking white paper to

  • Establish key applications and identify overlaps between domains scientific applications for resource-constrained ML and how they complement each other
  • (Brief) Review of ML techniques and compute technology for the domain scientist
  • Timeline: (tentatively) Feb 15, 2021 - details to follow in Indico email just after workshop

7 of 8

Scientific domain

applications

Machine learning techniques

Compute elements & systems

Domain-�inspired ML

Science�data�pipelines

ML-specific�systems

8 of 8

Scientific and Industrial

applications

Machine learning techniques

Computational systems and software

Domain-�specific ML

Data

analysis

systems

ML-specific�systems