ETD

• b-tagging
• cms
• hep
• higgs
• lhc
• machine learning

The ability to accurately identify and classify heavy-flavour jets produced in high-energy particle collisions is crucial for many analyses in CMS. In this thesis we introduce the ParticleEdge network, a Graph Neural Network-based b-tagging algorithm that showcases superior performance compared to the previous-generation algorithms utilized in CMS. Moreover, additional research was undertaken to assess the implications of improved b-tagging techniques on the analysis of the Higgs boson decay to b-quarks.