Parts | Days | Selection | Search | Updates | Downloads | Help
T: Fachverband Teilchenphysik
T 27: Searches II
T 27.1: Talk
Tuesday, March 21, 2023, 17:00–17:15, HSZ/0403
CMS Dijet Anomaly Search with Substructure — Gregor Kasieczka, •Louis Moureaux, Tobias Quadfasel, and Manuel Sommerhalder — Institut für Experimentalphysik, Universität Hamburg
The extensive searches for physics beyond the Standard Model carried out at the LHC have so far yielded no positive result, despite the very large number of models that have been tested. This motivates the use of techniques based on machine learning that, unlike common search strategies, are capable of dynamically adjusting the event selection to the observed data. These ``anomaly detection'' methods are expected to feature broad coverage of potential new physics signatures and can thus fill the gaps between searches dedicated to specific models.
We present the application of such an anomaly detection method, CATHODE, in a search for resonant dijet events using substructure observables with the CMS experiment. CATHODE combines density estimation and weak supervision techniques to detect anomalous events in a signal region, interpolating the background from sidebands to achieve nearly optimal classification performance.