Luigi Favaro, When physics meets machine learning: new tools for a new era
PSK 3rd floor, seminar room 1
Title:
When physics meets machine learning: new tools for a new era
Abstract:
The interplay between physics and machine learning is opening a new era in how we simulate and analyze collider events. In this talk, I will show how machine learning is affecting the collider simulation chain by replacing expensive components with precise surrogate models that maintain accuracy across the full phase space. Beyond speed, the most powerful gains come from integrating physical knowledge directly into model architectures. Whether through physics-motivated parameterizations or by directly embedding known symmetries into the architecture, we obtain models that are more robust, data-efficient, and physically consistent. Throughout, I will argue that physicists bring irreplaceable expertise to this effort, and that the dialogue between domain knowledge and modern machine learning is what will ultimately maximize the discovery potential of experiments like the LHC.
Zoom:
[TBA]