Bonn 2020 – wissenschaftliches Programm

Die DPG-Frühjahrstagung in Bonn musste abgesagt werden! Lesen Sie mehr ...

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

AKPIK: Arbeitskreis Physik, moderne Informationstechnologie und Künstliche Intelligenz

AKPIK 2: AKPIK II

AKPIK 2.5: Vortrag

Donnerstag, 2. April 2020, 17:30–17:45, H-HS XII

Adversarial Neural Network for ttH — •José Manuel Clavijo Columbié, Judith Katzy, and Paul Glaysher for the ATLAS collaboration — DESY, Notkestr.85, 22607 Hamburg

Measurements of ttH in the H->bb decay channel are attractive since this is the most frequent Higgs decay channel. However, it suffers from large tt+bb background which is usually separated by the use of classification machine learning algorithms trained on Monte Carlo simulated events. The largest uncertainty of the measurements usually stems from training bias towards a specific MC model. We apply adversarial domain adaptation to train a neural network that simultaneously classifies signal versus background events while minimizing the difference of the classifier response to two alternative background MC models by adding a discriminator with a gradient reversal layer.