MSc. Monika Alexandra Wüst and MSc. Luís Miguel Marques Lourenço
Julius-Maximilians-Universität Würzburg
T a l k : 26. June 2025
From classifier to bump hunt: methods in resonant anomaly detection
Resonant anomaly detection (RAD) methods have been developed to enhance the sensitivity of traditional bump hunt searches. The main components of these methods are the use of a background template to train a classifier, which assigns an anomaly score to events, and a subsequent statistical analysis of the data points that survive a selection based on this score. After introducing the general idea behind RAD, we present two methods developed in this context - CWoLA and CATHODE - which we use to illustrate the application of ML tools to produce both the classifier and the background template.