Login
Seminari di Sezione

From the Higgs to the Unknown: How AI is Reshaping the Search for New Physics

by Loukas Gouskos (Brown University)

→ Europe/Rome
Edificio C - Sala 131 (INFN - Pisa)

Edificio C - Sala 131

INFN - Pisa
Description

Abstract:

Artificial Intelligence has become indispensable to particle physics at the LHC, underpinning the reconstruction and tagging algorithms behind several analyses, enabling real-time inference in the trigger, and sharpening searches for subtle, correlated deviations from the Standard Model across very large datasets. I will begin with this experimental landscape, using Higgs physics and jet tagging as concrete examples of how AI has reshaped what we can measure and how far the precision frontier can be pushed. Despite decades of data, no direct evidence of physics beyond the Standard Model has emerged. Part of the difficulty is that the space of possible theories is vast and we can explore only a few models at a time. I will argue that the bottleneck is shifting from data to theory, and present Albert, a neuro-symbolic framework that constructs quantum field theories directly from data. Albert generates symmetries, particle content, and interactions under a formal grammar, so that every candidate is a consistent Lagrangian by construction, and uses reinforcement learning to score theories against experimental observables. As a proof of concept, a compact transformer trained only on legacy LEP data, which contains no direct evidence of the top quark, rediscovers it from precision electroweak observables alone. I will close with an outlook on how AI may transform the full arc of discovery, from measurement to theory.

 

Coordinate ZOOM

lg_higgs2unknown_infn_2026.pdf
lg_higgs2unknown_infn_2026.pptx