Résumé:
The Higgs boson discovery took place as late as 2012, after decades of long search. Its detection
was challenging, since it is produced from very rare processes with tiny cross sections around
2:2fb to be covered with huge amounts of background noise, to instantly into new constituents.
In this thesis, the main objective is to enhance that tiny signal, meaning nding the region
in the phase space where the Higgs signal is more dominant. Thanks to datasets provided
byATLAS opendata, we are able to nd an optimal model architecture to boost the Higgs
signal, by training di erent optimization techniques, where we adjust their parameters and add
non correlated variables. To do so we start by using traditional techniques, but since they are
not that e ective in increasing the signal, we are obligated to use machine learning methods
in the form of arti cial neural networks and boosted decision trees algorithms as they can
adjust several parameters simultaneously. After multiple training sessions, we achieve optimal
architectures to enhance the Higgs signal from the real data with reasonable value.
Keywords:
Higgs boson, Machine Learning, Arti cial Neural Networks, Boosted Decision Trees.