Hamburg 2016 – scientific programme

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

T: Fachverband Teilchenphysik

T 67: Higgs-Boson (assoziierte Produktion) III

T 67.3: Talk

Wednesday, March 2, 2016, 17:15–17:30, VMP5 HS B1

Categorization of the Processes Contributing to ttH(H→bb) Using Deep Neural Networks with the CMS Experiment — •Yannik Rath, Martin Erdmann, Benjamin Fischer, Robert Fischer, Fabian Heidemann, Thorben Quast, and Marcel Rieger — III. Physikalisches Institut A, RWTH Aachen University

In ttH(H→bb) analyses, event categorization is introduced to simultaneously constrain signal and background processes. A common procedure is to categorize events according to both their jet and b-tag multiplicities.

The separation power of this approach is limited by the b-tagging efficiency. Especially ttH(H→bb) events with their high b-tag multiplicities suffer from migrations to background categories.

In this presentation, we explore deep neural networks (DNNs) as a method of categorizing events according to their jet multiplicity and a DNN event class hypothesis. DNNs have the advantage of being able to learn discriminating features from low level variables, e.g. kinematic properties, and are naturally suited for multiclass classification problems. We compare the ttH signal separation achieved with the DNN method with that of a common categorization approach.