Heidelberg 2022 – wissenschaftliches Programm

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

T: Fachverband Teilchenphysik

T 26: Data Analysis, Information Technology and Artificial Intelligence

T 26.9: Vortrag

Montag, 21. März 2022, 18:15–18:30, T-H39

Improvement of the Jet-Parton Assignment in ttH(bb) Events using Symmetry-Preserving Attention Networks — •Daniel Bahner, Andrea Knue, and Gregor Herten — Albert-Ludwigs-University, Freiburg, Germany

The associated production of a Higgs boson and a top-quark pair allows for a direct measurement of the top-Higgs Yukawa coupling, which can be sensitive to Beyond Standard Model physics. In the studies presented, the process of interest is the semileptonic decay of the tt-pair accompanied by a bb-pair resulting from the most prominent Higgs decay. In this topology, at least four b-jets and two light jets are expected. This Higgs decay channel suffers from irreducible background due to tt+bb production. Furthermore, the full reconstruction of this final state proves difficult because of the ambiguities in assigning the jets to their original parton, which is called combinatorial background.

In the latest publication, a Boosted Decision Tree was used for the jet-parton assignment. In the studies presented in this talk, a novel Symmetry-Preserving Attention Network is exploited (suggested in arXiv:2106.03898). The training was performed and evaluated on two different samples: In the first sample the full detector simulation with Geant4 and the new ATLAS b-tagging algorithm DL1r was used and in the second sample the Delphes framework was used. The performances of the networks and possible future improvements will be presented.