Göttingen 2025 – scientific programme

Parts | Days | Selection | Search | Updates | Downloads | Help

T: Fachverband Teilchenphysik

T 2: Higgs Physics I (HH and trilinear coupling)

T 2.3: Talk

Monday, March 31, 2025, 17:15–17:30, ZHG104

Phase Space Optimization for the bbτ⁻τ⁺ Di-Higgs Analysis using Machine Learning with the CMS Experiment — Ana Andrade, •Anas Haddad, Philip Keicher, Tobias Kramer, Nathan Prouvost, Marcel Rieger, Peter Schleper, and Bogdan Wiederspan — Institute for experimental physics, University of Hamburg, Hamburg, Germany

This year marks the twelfth anniversary of the Higgs boson discovery. Yet, many of its properties and couplings remain unexplored. Particularly interesting are the couplings producing a Di-Higgs system in the final state, which are modulated as κ_λ and κ_2V in the κ-framework and pose a significant challenge for analyses due to the extremely low cross-sections of their production processes.

Since an efficient usage of the available data is crucial in such analyses, the selection is an important part and decisive for all following analysis steps and resulting measurements. However, one is always confronted with the dilemma of having to trade off higher event statistics for large background contamination in the selected phase space, or vice versa.

This study aims to move away from a fully cut-based selection, usually based on a certain topology, towards a more data-driven approach. The latter utilizes a NN on top of a loose preselection with the goal of optimizing the event selection in the search for Di-Higgs production in the bbτ⁻τ⁺ channel and enhancing the sensitivity of this analysis.

Keywords: Event Selection; Machine Learning; Di-Higgs; bbtautau