SMuK 2023 – wissenschaftliches Programm
Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
T: Fachverband Teilchenphysik
T 34: ML Methods II
T 34.2: Vortrag
Dienstag, 21. März 2023, 17:15–17:30, HSZ/0405
Identification of bb-Jets Using a Deep-Sets-Based Flavour-Tagging Algorithm with the ATLAS Experiment — •Joschka Birk1,2, A. Froch1, M. Guth3, and A. Knue1 — 1University of Freiburg — 2University of Hamburg — 3University of Geneva
Jets that contain two b-hadrons (bb-jets) are usually not considered as an individual target class in flavour-tagging algorithms. Instead, these jets are included in an inclusive b-jet category which consists of single-b jets and bb-jets, making these two types of jets indistinguishable when they are processed with such an algorithm.
While this is sufficient for most physics analyses, an explicit identification of bb-jets could be promising for analyses like the search for the ttH(→ bb) signal, which suffers from the large irreducible tt+bb background. This irreducible background contains the same final-state particles as the signal, including four b-quarks. In the background process, a radiated gluon can split into a b-quark pair, which might be contained in one single jet. In order to improve the rejection of these particular background events, the ATLAS DL1d algorithm, which is the b-tagging algorithm designed for ATLAS Run 3 analyses, is extended with an additional output class dedicated to bb-jets (bb-DL1d).
By applying a cut in a two-dimensional discriminant plane, bb-DL1d provides a proof-of-concept for a flavour-tagging algorithm that is capable of both inclusive b-tagging and bb-jet identification. The design of the bb-DL1d algorithm and its most important, Deep-Sets-based, low-level tagger bb-DIPS are discussed in this talk. Futhermore, performance studies for both algorithms are shown.