$H\rightarrow\tau\tau$ cross section measurements using machine learning and its interpretation as global couplings with the ATLAS detector

The Brout-Englert-Higgs mechanism provides a way for particles in the Standard Model to acquire mass. Specifically for fermions, Yukawa interactions with the Higgs boson are introduced to create the mass terms after spontaneous symmetry breaking. This thesis describes the measurement of the Higgs boson production cross section with the $H\rightarrow\tau\tau$ decay channel with the ATLAS detector. The analysis uses events from proton–proton collisions at a center-of-mass energy of $\sqrt{s}$ = 13 TeV. The analysis is performed on the full Run 2 dataset corresponding to an integrated luminosity of 139 fb${}^{−1}$. Cross section measurements are presented for the total cross section, for the gluon fusion and vector boson fusion production modes, for Higgs boson production in association with a vector boson or a pair of top quarks, and for nine fiducial regions of phase space in the Simplified Template Cross Section framework. The precision of the analysis is enhanced using Boosted Decision Trees. As an alternative, a deep multiclass neural network is trained, optimized, and evaluated. The multiclass approach achieves an approximately 8 % reduction in the uncertainty of the vector boson fusion cross section measurement. The measured product of total cross section and branching ratio is $0.93^{+0.13}_{-0.12}$ times the Standard Model expectation. A partially unblinded fit is used to validate the fit model and the assumptions on correlations between parameters in the fit. The results of the $H\rightarrow\tau\tau$ analysis are combined with the results from other Higgs boson channels. The combination is interpreted with global coupling parameters in the $\kappa$-framework. Several models have been evaluated with resolved or effective loop-induced couplings, and assumptions on contributions from beyond the Standard Model to the total Higgs boson decay widths. The improvements in the $H\rightarrow\tau\tau$ analysis compared to previous iterations are directly reflected in a reduction of the uncertainties on the coupling modifiers. All observed couplings are in agreement with the Standard Model.