Speaker
Description
Machine Learning for trigger/DAQ was among the first applications of ML in particle physics in the late 1980s. While great strides have been made since then in ML for physics, in analysis and reconstruction, ML at the edge has not been as prominent. Trigger/DAQ systems are a latency bound, custom electronics arena presenting challenging technical requirements for ML inference. Now, following advances in trigger and DAQ capabilities, and with significant community effort into tooling like hls4ml, ML is being deployed steadily closer to the detector. In the past few years, ML models have actively been making decisions in the detector readout chain, and analysis of ML-triggered data is underway. This talk will present some of the recent successes in ML for trigger/DAQ at the LHC and elsewhere, and look towards the next few years where ML for triggering is expected to be widely adopted at the HL-LHC.