Application of Machine Learning for the Charged Higgs Boson Search Using ATLAS Data

The discovery of the Higgs boson (2012) motivated scientists searching for charged Higgs bosons. The presence of a charged Higgs boson is predicted by many theories that describe an extended Standard Model, with several different Higgs bosons, called the "extended Higgs sector". Neural networks (NN) have recently been a big trend for solving classification, detection, and segmentation tasks. The advantage of NN is their ability to learn complex relationships hidden in data without any restrictions on the input data. The aim of this thesis is to separate the Signal process tbH+ from the Background processes. In this thesis, two NN architectures were tested: Multi-Layer Perceptron (MLP) and TabNet. A good separation of Signal and Background was obtained as a function of the charged Higgs boson mass.