Bonn 2020 – scientific programme

The DPG Spring Meeting in Bonn had to be cancelled! Read more ...

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

T: Fachverband Teilchenphysik

T 47: Neural networks and systematic uncertainties

T 47.4: Talk

Wednesday, April 1, 2020, 17:15–17:30, H-HS IV

Reduction of systematic uncertainties with adversarial neural networks in scope of the ttH(bb) analysis at CMS — •Simon Ehnle, Ulrich Husemann, Philip Keicher, Matthias Schröder, and Sebastian Wieland — Institut für Experimentelle Teilchenphysik (ETP), Karlsruher Institut für Technologie (KIT)

The production of top quark-antiquark pairs in association with the Higgs boson allows a direct measurement of the top-Higgs Yukawa coupling. To compensate the small cross section, the Higgs boson decay into a bottom quark-antiquark pair (ttH(bb)), which has the largest branching ratio, is investigated. Multivariate analysis methods are used to separate signal from background.

A major background in this channel is the top quark-antiquark pair production in association with a bottom quark-antiquark pair. This process is hard to model and different simulation approaches with different uncertainties exist. The classifying neural networks have the potential to get robust against these differences by using adversarial neural networks, whereby two neural networks compete against each other in a zero-sum game.

In this presentation, the approach of reducing systematic uncertainties with adversarial neural networks is studied in scope of the ttH(bb) analysis in the semileptonic channel at CMS.