Learning new physics efficiently with nonparametric methods

• Published in: The European physical journal. C, Particles and fields Vol. 82; no. 10; p. 879
• Main Authors: Letizia, Marco, Losapio, Gianvito, Rando, Marco, Grosso, Gaia, Wulzer, Andrea, Pierini, Maurizio, Zanetti, Marco, Rosasco, Lorenzo
• Format: Journal Article
• Language: English
• Published: Berlin/Heidelberg Springer Berlin Heidelberg 01.10.2022; Springer; Springer Nature B.V; Springer Verlag (Germany); SpringerOpen
• Subjects: Algorithms; Analysis; Astronomy; Astrophysics and Cosmology; Computer Science; Continuity (mathematics); Data mining; Datasets; Elementary Particles; Experiments; Hadrons; Heavy Ions; High energy physics; High Energy Physics - Phenomenology; Hypotheses; Hypothesis testing; Kernel functions; Likelihood ratio; Machine Learning; Measurement Science and Instrumentation; Methods; Neural networks; Nonparametric statistics; Nuclear Energy; Nuclear Physics; Physics; Physics and Astronomy; Quantum Field Theories; Quantum Field Theory; Random variables; Regular - Experimental Physics; Regular Article - Experimental Physics; String Theory; anomaly detection; goodness-of-fit; nonparametric methods; efficient machine learning; two-sample test; model-independent
• ISSN: 1434-6052; 1434-6044; 1434-6052
• DOI: 10.1140/epjc/s10052-022-10830-y

We present a machine learning approach for model-independent new physics searches. The corresponding algorithm is powered by recent large-scale implementations of kernel methods, nonparametric learning algorithms that can approximate any continuous function given enough data. Based on the original proposal by D’Agnolo and Wulzer (Phys Rev D 99(1):015014, 2019, arXiv:1806.02350 [hep-ph]), the model evaluates the compatibility between experimental data and a reference model, by implementing a hypothesis testing procedure based on the likelihood ratio. Model-independence is enforced by avoiding any prior assumption about the presence or shape of new physics components in the measurements. We show that our approach has dramatic advantages compared to neural network implementations in terms of training times and computational resources, while maintaining comparable performances. In particular, we conduct our tests on higher dimensional datasets, a step forward with respect to previous studies.