Imposing exclusion limits on new physics with machine-learned likelihoods

Abstract

Machine-Learned Likelihood (MLL) is a method that, by combining modern machine-learning techniques with likelihood-based inference tests, allows estimating the experimental sensitivity of high-dimensional data sets. Here we extend the MLL method by including the exclusion hypothesis tests and study it first on a toy model of multivariate Gaussian distributions, where the true probability distribution functions are known. We then apply it to a case of interest in the search for new physics at the LHC, in which a ′ boson decays into lepton pairs, comparing the performance of MLL for estimating 95% CL exclusion limits with respect to the prospects reported by ATLAS at 14 TeV with a luminosity of 3 ab−1.