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

by Mikael Kuusela (Carnegie Mellon University (US))

Wednesday 27 Apr 2022, 16:00 → 17:30 Europe/Zurich

N.B. This seminar is part of the PHYSTAT-Anomalies Workshop. The rest of the Workshop is on 24th and 25th May. Further information, including the timetable and registration details are at the Indico page: https://indico.cern.ch/event/1138933/ 

 

A central goal in experimental high energy physics is to detect new signals that appear as deviations from known Standard
Model physics in high-dimensional particle physics data. To do this, one can determine whether there is any statistically
significant difference between the distribution of Standard Model background samples and the distribution of the experimental observations, which are a mixture of the background and a potential new signal. Traditionally, one also assumes access to a sample from a model for the hypothesized signal distribution. Here we instead investigate 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 in the experimental data. 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). Additionally, we propose a method for estimating the signal strength parameter and explore active subspace methods to interpret the fitted semi-supervised classifier in order to understand the properties of the detected signal. We investigate the performance of the methods on a data set related to the search for the Higgs boson at the Large Hadron Collider. We demonstrate that the semi-supervised tests have power competitive with the classical supervised methods for a well-specified signal, but much higher power for an unexpected signal which might be entirely missed by the supervised tests.

Joint work with Purvasha Chakravarti, Jing Lei and Larry Wasserman.

Mikael Kuusela is a professor of Statistics and Data Science at Carnegie Mellon University. His research focuses on methods for analysing large and complex data sets in the physical sciences, including unfolding and statistical learning problems in high-energy physics. He has been a member of CMS since 2010, and has extensive experience explaining statistical issues to Particle Physicists.
 

 

 

Kuusela_Phystat_Apr_2022.pdf

video1858869988.mp4