ECE Graduate Seminar

  • Remote Access Enabled - Zoom
  • Virtual Presentation
  • Assistant Professor
  • Department of Physics, University of California, San Diego
  • MenberCMS collaboration at CERN

Real-time AI in particle physics

Experimental high energy physics is entering an era of unprecedented data rates. The vast majority of the raw data in our experiments are immediately discarded by our real-time trigger systems because of storage and computational limitations. While most of these data are background, we may inadvertently be throwing away precious signals of new physics. One way to expand our discovery potential is by enhancing our real-time on-detector and trigger-level processing capabilities using machine learning (ML). I will discuss applications and opportunities for ML in real-time systems in particle physics, focusing on how we are embedding ML algorithms in systems with FPGAs and ASICs for a variety of use-cases. I will review essential ideas for designing and optimizing efficient algorithms in hardware and emerging tool flows to accelerate algorithm development. I will then explore a few examples spanning different applications, including more sophisticated trigger algorithms, front-end data compression in radiation-hard environments, and controls of particle accelerators.

Javier Duarte is an Assistant Professor of Physics at the University of California San Diego and a member of the CMS experiment at the CERN Large Hadron Collider (LHC). Before joining UCSD, he was a Lederman postdoctoral fellow at Fermilab. He received his Ph.D. in Physics at Caltech and his B.S. in Physics and Mathematics at MIT. His main research interests are high-momentum Higgs boson measurements, searches for exotic new physics, real-time AI on FPGAs, and geometric deep learning for particle physics. Prof. Duarte has received a DOE Early Career Award for real-time AI for particle reconstruction and Higgs physics.

Zoom Participation. See announcement.