Recent developments in machine learning (ML) are rapidly changing physics reconstruction algorithms. These are leading to better-performing algorithms with faster computation times. In this workshop, we investigate both hardware and algorithmic ML approaches to speed up inference in data acquisition, focusing on the HL-LHC trigger upgrade, along with the potential to further compress the event information to produce smaller data samples.
The goal of this workshop is to develop cross-experiment collaborations to work on common problems that will be faced by HL-LHC experiments. Beyond this, we hope to build collaborations between the HEP and the CS communities. The irregular 3-D geometries of HEP detectors, their heterogeneity, and the extremely small latencies provide unique and interesting data science challenges. Conversely, we have excellent models of how particles behave in our detectors, and excellent simulations of these detectors, which makes obtaining vast training samples possible. By building HEP-CS collaborations, we hope that HEP data sets can be used for cutting-edge ML research, providing benefits to both communities.