Transforming Particle Theory

by Tilman Plehn

Wednesday 11 Mar 2026, 11:00 → 12:00 Europe/Zurich

222/R-001 (CERN)

Machine learning is not only transforming our lives, but also the way we do particle theory. I will describe a few established ML-applications in particle theory, leading to symmetries, uncertainties, and eventually physics-specific representation learning. I will then give examples how we can use learned representations to improve network performance and understand what networks do. Finally, I will briefly show how agent-based ML-simulations can re-define the way we do physics.

Bio:

Tilman Plehn is a leading figure in theoretical particle physics who was among the earliest researchers to recognize and embrace the transformative potential of deep learning in the field. Long before deep learning became mainstream in high-energy physics, he championed its use for tackling complex problems such as event classification, parameter inference, and uncertainty quantification. He consistently took a pioneering role, not only applying modern AI methods but also helping shape how the community thinks about them. In particular, he was a driving force behind the introduction and development of normalizing flows for particle-physics applications, opening new avenues for fast simulation, likelihood-free inference, and generative modeling. Through this sustained leadership across multiple fronts, Plehn has played a central role in bridging modern machine learning and fundamental physics.

 

Coffee will be served at 10h30