Siam Physics Congress 2022 (SPC2022)

Search for BSM Higgs bosons using Machine Learning techniques

S4 High Energy and Particle Physics

23 Jun 2022, 16:00

15m

QUARTZ

Board: O-S4-7

Oral Presentation High Energy and Particle Physics S4 High Energy and Particle Physics

Speakers

Jinna Waiwattana (Mahidol Wittayanusorn School) Natha Pitakkultorn (Mahidol Wittayanusorn School) Pakorn Saksirimontri (Mahidol Wittayanusorn School)

Description

Although the discovery of the 125 GeV Higgs boson has confirmed the Higgs mechanism of the Standard Model (SM), many theories beyond the Standard Model (BSM) has been introduced to solve several phenomena not explained by the SM. One example is the Two Higgs Doublet Models (2HDM), the simplest extension of the SM Higgs sector, that predicted the existence of additional Higgs bosons. In this study, we investigate the use of supervised learning of machine learning method which is hypothesized to be more effective than traditional cut-based method, to search for BSM Higgs bosons decaying into bottom-antibottom quark pairs ($H\to bb$), the dominant Higgs boson decay channel. We train machine learning models to classify whether events detected by the detector belongs to $H\to bb$ process (signal) or multijet process (background) based on simulated proton-proton collisions within the Compact Muon Solenoid (CMS) detector. The evaluation matrix is calculated to compare the classification efficiency of models using Neural Networks (NN) and Tree based models. The results show that machine learning models can classify signal and background processes with significant improvement and can be used as signal-background discriminator for further statistical analyses searching for BSM Higgs bosons.