Model-Independent Detection of New Physics Signals Using Interpretable Semi-Supervised Classifier Tests

• Activity Number: 28 - Computation, Design, and Quality Assurance of Physical Science and Engineering Applications
• Type: Contributed
• Date/Time: Sunday, August 8, 2021 : 1:30 PM to 3:20 PM
• Sponsor: Section on Physical and Engineering Sciences
• Author(s): Purvasha Chakravarti* and Mikael Kuusela and Jing Lei and Larry Wasserman
• Companies: Imperial College London and Carnegie Mellon University and Carnegie Mellon University and Carnegie Mellon University
• Keywords: collective anomaly detection; active subspace; mixture proportion estimation; signal strength estimation; likelihood ratio test; high-dimensional two-sample testing

Abstract

We address a central goal in experimental high energy physics -- to search for new signals that appear as deviations from known Standard Model physics in high-dimensional particle physics data. To do this, we determine whether there is any statistically significant difference between the distribution of Standard Model background samples and the distribution of the experimental data, which are a mixture of the background and a potential new signal. Unlike most previous work, we use a model-independent method that does not make any assumptions about the signal and uses a semi-supervised classifier to detect the presence of the signal. We construct three test statistics using the classifier: an estimated likelihood ratio test (LRT) statistic, a test based on the area under the ROC curve (AUC), and a test based on the misclassification error (MCE). We also propose a method for estimating the signal strength parameter and explore active subspace methods to interpret the proposed semi-supervised classifier in order to understand the properties of the detected signal. We show the competitive performance of the methods on data related to the search for the Higgs boson at the LHC at CERN.