Studien zu Top-Quark und Photon Kopplung mit dem ATLAS-Experiment

Neural Networks and other machine learning algorithms enjoy an ever-expanding usage in particle and high energy physics. Primarily, they are used as binary classifiers to discriminate between signal and background contributions. The thesis deals with the usage of a Neural Network to discriminate between prompt photons, which will be radiated off top quark pairs, and the contribution of hadron fakes being a dominant background contribution to get sensitive to the tγ coupling for a centreof-mass energy √ s = 13 TeV at Cern. The structure of the Atlas detector is important for this study. So-called shower shapes will be implemented in the used Neural Network to eliminate systematic uncertainties in a best possible way. As it turns out, they will be used to get the desired separation between prompt photons and hadron fakes.