Measurement of the $t\bar{t}+ Z$ Cross Section in the Four Lepton Channel Using a Deep Neural Network

The production of top-quark pairs in association with a Z boson offers the opportunity to study the coupling of the Z boson to the top quark. This coupling is of interest because it is sensitive to the third component of the weak isospin and the hypercharge of the top quark. A measurement of the coupling probes the prediction of these parameters of the Standard Model of elementary particle physics (SM). This thesis presents studies of the final states containing four electrically charged lep- tons. This channel is particularly pure but such events are rare. In order to increase the statistical precision, a deep neural network (DNN) was trained to perform the clas- sification of events. The preselection is chosen to be loose to increase the acceptance of four-lepton events. Then, the separation of background and signal events is performed by the DNN. This way the statistical uncertainty of the cross section measurement is reduced by 7% to 35% depending on the used configuration for the measurement compared to a “cut and count” approach. Within this thesis also the Z boson reconstruction is performed by a DNN instead of using a simple Z mass window based approach with the aim to increase the reconstruction efficiency. The reconstruction is required in order to study the coupling.