Please join us for a UW Data Science Seminar event on Wednesday, May 31st from 4:30 to 5:20 p.m. PST. The seminar will feature Elham E. Khoda, Postdoctoral Scholar at the UW Department of Physics.
“Fast Machine Learning on FPGAs for particle physics applications”
Abstract: In particle physics, we are experiencing a very high raw data rate at the Large Hadron Collider (LHC), where the protons collide at a 40 MHZ rate. It is impossible to read out and store all the data at this high rate. So, the particle detectors around the LHC ring use an electronic hardware “trigger” system to select potentially interesting particle collisions for further analysis. Currently, one out of 400 proton-proton collision events passes the hardware trigger. As the collision rate will increase by 5-7 times in the future alternative algorithms, such as ML, can be used for fast and accurate decisions.
In this talk, I will highlight the potential applications of ML for hardware (ASIC or FPGA) triggers. I will discuss a method to implement the ML algorithms on an FPGA using the hls4ml software package. hls4ml is a user-friendly software based on High-Level Synthesis (HLS) designed to deploy neural network architectures on FPGAs. I will highlight my recent work on recursive neural networks (RNN)-based and Transformer-based algorithms for trigger applications.
Biography: Elham E Khoda is a UW particle physics postdoc in the EPE group working with Prof. Shih-Chieh Hsu on new particle searches and Machine Learning algorithms for particle physics. He completed his Ph.D. in particle physics at the University of British Columbia, Vancouver, Canada, on “Searches for new high-mass resonances in top-antitop and di-electron final states using the ATLAS detector”. He is interested in developing ML algorithms to solve particle physics challenges. He is a major contributor to EPE’s activities toward data-driven discovery with accelerated AI algorithms. He is working on accelerating ML inference with coprocessors like GPUs and FPGAs.
The UW Data Science Seminar is an annual lecture series at the University of Washington that hosts scholars working across applied areas of data science, such as the sciences, engineering, humanities and arts along with methodological areas in data science, such as computer science, applied math and statistics. Our presenters come from all domain fields and include occasional external speakers from regional partners, governmental agencies and industry.
The 2022-2023 seminars will be virtual, and are free and open to the public.