Anomaly Detection Using Isolation Forests in Searches for Inelastic Dark Matter with a Dark Higgs at Belle II

Anomaly detection as a method enables the study of new physical phenomena. This thesis studies the search for Dark Higgs signals given by the Inelastic Dark Matter with a Dark Higgs model. Using MC simulations and the Isolation Forests approach, collisions of electron-positron pairs at Belle II are analyzed to detect deviations in the event distributions. Signals of potential signatures in the detector indicating Dark Higgs candidates are investigated. The study provides a comprehensive insight into the discrimination of different Standard Model processes as well as the use of input feature information from the collision processes. In addition, the sensitivity of anomaly detection is evaluated by comparing the Isolation Forest with the Autoencoder approach.