Improving $t\overline{t}H$ Detection in ATLAS Experiment Using Machine Learning Techniques

Collision data are recorded at the rate of $40MHz$ in the Large Hadron Collider (LHC) with over $60 TB$ of data created every second, which contributes to over $10 GB$ of data being permanently stored in various data centers after initial triggering. Analysing this huge amount of data is challenging, since traditional ways of data analysis are too slow. Luckily, with the ever-advancing computing power, machine learning techniques are now applied to a variety of tasks. Therefore, as proposed by my supervisor, Dr. Nello Bruscino, I tried to use two different machine learning algorithms, Multi-Layer Perception (MLP), or I later referred as "Ordinary Neural Network", and Graph Neural Network (GNN) to help finding Higgs boson created in proton-proton collisions in associate with a couple of top-antitop quarks.