The Inverse Bagging Algorithm: Anomaly Detection by Inverse Bootstrap Aggregating
This work addresses anomaly detection in high-energy physics, specifically for searches at hadronic colliders like the LHC, but it appears incremental as it builds on existing bootstrap and bagging techniques.
The paper tackles the problem of anomaly detection in datasets with a well-modeled process and an unknown process, aiming to manipulate the unknown fraction without altering the kinematic distributions of the well-modeled one. It introduces the Inverse Bagging Algorithm, which uses bootstrap techniques to identify sub-samples rich in the well-modeled process and classifies events based on frequency, showing comparisons with general MVA algorithms and asymptotic properties using a public domain dataset modeling LHC physics searches.
For data sets populated by a very well modeled process and by another process of unknown probability density function (PDF), a desired feature when manipulating the fraction of the unknown process (either for enhancing it or suppressing it) consists in avoiding to modify the kinematic distributions of the well modeled one. A bootstrap technique is used to identify sub-samples rich in the well modeled process, and classify each event according to the frequency of it being part of such sub-samples. Comparisons with general MVA algorithms will be shown, as well as a study of the asymptotic properties of the method, making use of a public domain data set that models a typical search for new physics as performed at hadronic colliders such as the Large Hadron Collider (LHC).