Karlsruhe 2024 – scientific programme

Parts | Days | Selection | Search | Updates | Downloads | Help

T: Fachverband Teilchenphysik

T 71: Higgs 3 (coupling to b and c quarks)

T 71.2: Talk

Wednesday, March 6, 2024, 16:15–16:30, Geb. 30.41: HS 2

Background Modelling using ML in ttH(bb) Final States — Steffen Korn, Arnulf Quadt, Chris Scheulen, and •Paul Wollenhaupt — II. Physikalisches Institut, Georg-August- Universität Göttingen, Germany

Systematic differences between simulated samples and measured data are challenging for many high-energy physics analyses. The machine learning field of domain translation provides a powerful framework for learning mappings that systematically correct the distribution of simulated samples. Analogous to the ABCD method, which extrapolates the absolute number of events in a signal region (SR) from the translations of yields in control regions (CRs), mappings from sample to data distributions are first learned in the CRs and then extrapolated to the blinded SR. This domain translation approach is used to improve the background modelling of ttH(→ bb). Specifically, regions are defined based on the number of jets and b-tagged jets. The distributions of the jet kinematic variables, which are systematically mis-modelled due to the NLO approximation of the top quark, are then extrapolated for events with at least six hadronic jets, three of which are b-tagged.

Keywords: ttH; Background Modelling; Higgs Phyiscs; Machine-Learning