Heidelberg 2022 – wissenschaftliches Programm
Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
T: Fachverband Teilchenphysik
T 35: Higgs Boson: Associated Production 1
T 35.5: Vortrag
Dienstag, 22. März 2022, 17:15–17:30, T-H20
Adversarial Machine Learning Methods for Modelling Uncertainty Reduction in the Bottom Anti-Bottom Higgs Decay Channel of Higgs-associated Top Quark Pair Production with ATLAS at 13 TeV — Arnulf Quadt, •Chris Scheulen, and Elizaveta Shabalina — II. Physikalisches Institut, Georg-August Universität Göttingen
The bottom anti-bottom Higgs decay channel of Higgs-associated top quark pair production offers direct access to measurements of the top Yukawa coupling and Higgs0.05em-p0.01emT differential cross section, which are sensitive to potential new physics. To incorporate improvements such as developments in b0.1em-tagging and event simulation, a legacy analysis of the ttH(H → bb) process in the full ATLAS Run 2 Dataset of L = 139 fb0.05em-1 is currently ongoing.
Modelling differences between Monte Carlo samples of the dominant tt + jets background process were found to be one of the most significant sources of uncertainty in previous analysis rounds. Along with investigating and mitigating the source of these modelling differences via generator studies of new tt + jets background simulation setups or improving event classification performance to decrease background contamination in signal regions, the usage of adversarial machine learning techniques to select robust features could decrease the impact of background modelling systematics on the fit performance. This talk will present ongoing efforts concerned with developing such adversarial machine learning approaches.