With the start of Run 3 data-taking, Machine Learning (ML) has become an integral part of real-time event selection systems at the LHC.
Through advancements in algorithm design, ML algorithms are capable of swiftly processing millions of collisions per second on field programmable gate arrays (FPGAs) and perform efficient reconstruction and decision making.
As we are approaching the High Luminosity phase of the LHC, the triggering systems must deal with unprecedented data rates and event complexity. This will increasingly require us to explore new and efficient ML architectures to ensure that physics quality is maintained.
In this seminar, we will discuss how real-time ML is used to process and filter the enormous amount of data in LHC experiments in order to improve physics acceptance. We will discuss state-of-the-art workflows for designing and deploying ultra fast ML algorithms on FPGA hardware, with a focus on the hls4ml and conifer libraries. Finally, we will explore a few example applications of real-time inference in particle physics experiments.
Thea Klaeboe Aarrestad is a Research Fellow at ETH Zürich and a member of the CMS Collaboration at CERN. Her research centers on how Machine Learning can be applied to particle physics problems, especially focusing on utilizing real-time ML for discovering new physics phenomena. She received her Ph.D. in Physics from the University of Zürich and spent two years as a Research Fellow at CERN before moving to Zürich as an SNSF Ambizione grantee.
Coffee will be served at 10:00.
M. Girone, M. Elsing, L. Moneta, M. Pierini