Bonn 2020 – wissenschaftliches Programm
Die DPG-Frühjahrstagung in Bonn musste abgesagt werden! Lesen Sie mehr ...
T 47.3: Vortrag
Mittwoch, 1. April 2020, 17:00–17:15, H-HS IV
Treating Uncertainties with Bayesian Neural Networks in a ttH Measurement — Ulrich Husemann, Philip Keicher, Matthias Schröder, and •Nikita Shadskiy — Institut für Experimentelle Teilchenphysik (ETP), Karlsruher Institut für Technologie (KIT)
In the Standard Model, fermions couple to the Higgs boson via a Yukawa coupling with a strength proportional to their mass. The top quark is the heaviest known fermion and, therefore, has the strongest coupling to the Higgs boson.
One of the processes to investigate this coupling is the associated tt+H production where the Higgs boson decays into a bb pair. This signal process has a much smaller cross section than the challenging background processes like tt+jets production. Especially tt+bb events are very signal-like. A common approach to separate this signal from the backgrounds is to use artificial neural networks.
Neural networks normally do not take into account uncertainties of the processes. Bayesian neural networks, however, use whole weight distributions instead of single weight values. In this talk it is investigated how this feature of Bayesian neural networks can be used to treat uncertainties in a ttH measurement.