19–23 May 2025
CERN
Europe/Zurich timezone

Hadronic tau identification in CMS applying domain adaptation techniques to a CNN

Not scheduled
20m
61/1-201 - Pas perdus - Not a meeting room - (CERN)

61/1-201 - Pas perdus - Not a meeting room -

CERN

10
Show room on map
Poster 1 ML for object identification and reconstruction Poster Session

Speaker

Olha Lavoryk (KIT - Karlsruhe Institute of Technology (DE))

Description

The CMS experiment has deployed for the Run 2 LHC data-taking period a Convolutional Neural Network architecture to identify hadronically decaying tau leptons against quark and gluon jets, electrons, and muons: the DeepTau algorithm. For the LHC Run 3, this algorithm saw an important upgrade with the introduction of domain adaptation techniques in order to improve its performance and achieve better modeling of the behavior in simulation with respect to recorded data. Further improvements to the network architectures are also discussed, together with its performance in early Run 3 data. This talk also provides an overview of other algorithms used within the CMS experiment for the identification of hadronically decaying taus in standard or rare topologies.

Would you like to be considered for an oral presentation? Yes

Authors

Aliaksei Raspiareza (Deutsches Elektronen-Synchrotron (DE)) Andrea Cardini (Universidad de Oviedo) Daniel Winterbottom (Imperial College (GB)) Irene Andreou (Imperial College (GB)) Konstantin Androsov (Texas A & M University (US)) Luca Guzzi (Universita & INFN, Milano-Bicocca (IT)) Lucas Russell (Imperial College (GB)) Michal Bluj (National Centre for Nuclear Research (PL)) Mykyta Shchedrolosiev (Deutsches Elektronen-Synchrotron (DE)) Oceane Poncet (Centre National de la Recherche Scientifique (FR)) Oleg Filatov Olha Lavoryk (KIT - Karlsruhe Institute of Technology (DE)) Paola Mastrapasqua (Universite Catholique de Louvain (UCL) (BE)) Saskia Falke (Centre National de la Recherche Scientifique (FR)) Stepan Zakharov (Deutsches Elektronen-Synchrotron (DE))

Presentation materials