Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe

T: Fachverband Teilchenphysik

T 67: Data, AI, Computing 5 (normalising flows)

T 67.1: Vortrag

Mittwoch, 6. März 2024, 16:00–16:15, Geb. 30.33: MTI

Applications of Normalizing Flows in High-Energy Particle Physics — •Lars Sowa, Roger Wolf, Markus Klute, and Günter Quast — Institute of Experimental Particle Physics (ETP), Karlsuhe Institute of Technology (KIT)

Normalizing flows (NFs) are neural networks that preserve probability when mapping probability density distributions from a given input to an arbitrary output space. They exhibit promising capabilities both, as surrogates for the fast generation of new samples as approximators of arbitrary probability density functions. These properties make them compelling in high-energy physics (HEP) applications. This work focuses on the application of NFs for recoil calibration, specifically with LHC Run-3 data taken with the CMS detector. Additionally, an NF-based b-jet regression is introduced, enabling the estimation of the per-jet energy resolution. The proposed methods showcase the versatility of NFs in HEP, serving as effective tools.

Keywords: Machine Learning; Normalizing Flows