Heidelberg 2022 – wissenschaftliches Programm

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

T: Fachverband Teilchenphysik

T 107: Data Analysis, Information Technology and Artificial Intelligence 5

T 107.7: Vortrag

Donnerstag, 24. März 2022, 17:45–18:00, T-H39

Usage of neural networks in photon identification in ATLAS — •Florian Kirfel — Physikalisches Institut der Universität Bonn

Precise photon identification is crucial for many ATLAS analyses. Currently, photons are selected using a set of cuts on calorimeter variables which characterise the shape of electromagnetic showers. These cuts were optimized using Monte Carlo simulations of photons and jets. Due to the simulations not being ideal, the selection efficiency must be corrected to match data. However, the measurement technique used to determine the identification efficiency in the data requires the hadronic activity around the photon candidate and the photon identification efficiency to be independent. In this work, neural networks are employed to improve over cut-based photon identification. In addition, they are constrained to keep the classification independent of the isolation using the distance correlation. This allows a simplified setup comparing to alternatives such as the adversarial neural network.