Speaker
Description
The application of AI, which ranges from task-specific models to large foundation models that are adaptable to different tasks and detectors, has been transformative for nuclear physics. However, realistic detector simulations and/or labeled data are needed for AI to be usable in experiments. Furthermore, given the significant resources required to train a massive foundation model, a baseline needs to be established to quantify the improvements these new algorithms can bring. In this presentation, I will focus on creating a realistic simulation of and establishing a baseline for the barrel imaging calorimeter (BEMC) in the ePIC detector at the EIC. I will show the effects included in the simulation, the baseline resolution and efficiency of the default non-AI clustering algorithm, as well as the performance of some specialized small-scale AI models on key benchmarks. I will also briefly mention how self-supervised learning (SSL) can be leveraged to train a foundation model and how it holds reasonable hope of delivering enhanced results on myriad tasks.