Aachen 2019 – scientific programme

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

T: Fachverband Teilchenphysik

T 77: Deep Learning III

T 77.9: Talk

Thursday, March 28, 2019, 18:00–18:15, H06

Multi-Class Boosted Object Tagger for Reclustered Jets at the ATLAS Experiment — •Elena Freundlich, Olaf Nackenhorst, Johannes Erdmann, and Kevin Kröninger — TU Dortmund, Experimentelle Physik IV

The identification of boosted objects is a key element of many analyses targeting the search for new physics at high energies. At the ATLAS experiment, a multi-class boosted object tagger for reclustered (RC) jets is developed as a novel tool to identify W/Z bosons, H bosons and t quarks unambiguously. A clear advantage of this tagger is the possibility to propagate the systematic uncertainties of small-R jets.

As input, RC jet properties and constituent information are used to train a deep neural network in order to obtain a four-dimensional output and differentiate between boosted objects and multijet events. The performance for each of the three signal classes W/Z, H and t can be adapted according to the needs of an analysis. Different studies about the performance of such a tagger are shown.