SMuK 2023 – scientific programme

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

T: Fachverband Teilchenphysik

T 79: Searches III

T 79.2: Talk

Wednesday, March 22, 2023, 17:45–18:00, HSZ/0403

Exploring extentions of MUSiC with Machine Learning techniques — •Ana Rita Alves Andrade, Thomas Hebbeker, Yannik Kaiser, Arnd Meyer, and Felipe Torres da Silva de Araujo — III. Physikalisches Institut A, RWTH Aachen University

MUSiC - Model Unspecific Search in CMS - is a model-independent search used in the CMS experiment, serving as a complementary approach to model-specific searches. Unlike the latter approach, MUSiC neither constrains the search phase-space nor is restricted to a specific final state. To this end, MUSiC employs, per set of final state multiplicity, an automated search for the most discrepant phase-space region, considering a defined p-value. We report results on exploring the implementation of the New Physics Learning from a Machine (NPLM) algorithm, a machine learning (ML) approach for new physics searches, applied to simulated MUSiC-like data as well as CMS data pre-processed by MUSiC. Sensitivities for the nominal MUSiC and the ML modified approach are discussed. Challenges to incorporate this or similar ML methods to the standard MUSiC procedure, are also considered.