Speaker
Saul Alonso-Monsalve
Description
Applying deep learning to neutrino physics offers powerful new capabilities for tasks such as event reconstruction, but it also exposes unique technical challenges. In this talk, I'll discuss lessons learned from the deep learning perspective: handling sparse, high-dimensional detector data; not limiting the capacity of our models; and ensuring model robustness and interpretability. These experiences highlight where standard deep learning practices break down in scientific contexts and how they can be adapted for reliable use in neutrino physics experiments.
