סמינר בחלקיקים: First ML-based algorithm for tau triggering in the ultra low latency domain of the Level-1 ATLAS trigger

Abstract:

The ATLAS experiment is producing vast amounts of data for physics analyses. With every upgrade it is becoming increasingly challenging to keep up with the data rates while retaining high event selection efficiency, pushing heuristics-based selection algorithms to their limit. Recent advancements in the field of hardware-based fast inference by machine learning models allow us to harness the power of ML in the realm of ultra-low latencies at the Level-1 ATLAS trigger.

BDTs are simple yet powerful ML algorithms with performance often at par with cutting-edge NN-based models. The structure of BDTs allows for a highly parallelized, low-latency implementation in FPGAs. I will describe the process of the development, implementation and commissioning of the first ever ML-based, BDT algorithm in the ATLAS Level-1 trigger for tau lepton identification from inception to production.

מארגן הסמינר: ד"ר מיכאל גלר