Erlangen 2026 – scientific programme

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

T: Fachverband Teilchenphysik

T 58: Searches/BSM III

T 58.7: Talk

Wednesday, March 18, 2026, 17:45–18:00, KH 02.018

Significance metrics for anomaly detection with the CATHODE method — •Tore von Schwartz, Louis Moureaux, Gregor Kasieczka, Chitrakshee Yede, and Kristian Warnholz — University of Hamburg, Hamburg, Germany

Despite an extensive search program at the LHC, no evidence for new physics has been found so far. Anomaly detection has emerged as a promising approach bridging generic searches and analyses targeting specific signals. CATHODE, a model-agnostic anomaly detection method, is designed to enhance resonant signals in the smoothly falling dijet invariant mass spectrum. We present an investigation of different strategies for extracting the signal significance after applying this method, with the aim of improving sensitivity to potential new physics signals while maintaining calibration.

Keywords: CATHODE; Anomaly Detection; Significance; Machine Learning; Resonant Searches