Powered by Indico v3.0.3
Abstract:
Real-time data processing is a central challenge in large-scale particle physics experiments, where tens of thousands of detector channels generate continuous data streams. Machine learning methods are increasingly integrated into trigger and acquisition systems to improve event identification efficiency and background rejection. Combined with edge computing and hardware acceleration, these approaches enable low-latency and high-consistency performance directly at the experimental site.
This seminar reviews recent progress in machine learning for real-time applications, with a focus on representative algorithmic strategies and hardware platforms such as FPGAs and GPUs. Case studies from the JUNO experiment illustrate how, under limited hardware resources, low latency and efficient classification can be successfully achieved, demonstrating the value of combining machine learning with hardware solutions in large-scale scientific experiments.
About the speaker:
Feng Gao received her Ph.D. in particle physics from RWTH Aachen University (Germany), where she focused on the design and production of intelligent photomultiplier tubes for the OSIRIS detector. She is currently a postdoctoral researcher at the Inter-University Institute for High Energies, ULB Brussels, working on the JUNO experiment’s trigger system, online event classification, and FPGA-based edge machine learning.
zoom会议号: 9316 3704 686
入会密码Password：250905