Deconvolution of the High Energy Particle Physics Data with Machine Learning

Abstract

A method for correcting smearing effects using machine learning technique is presented. Compared to the standard deconvolution approaches in high energy particle physics, the method can use more than one reconstructed variable to predict the value of unsmeared quantity on an event-by-event basis. In this particular study, deconvolution is interpreted as a classification problem, and neural networks (NN) are trained to deconvolute the Z boson invariant mass spectrum generated with MadGraph and pythia8 Monte Carlo event generators in order to prove the principle. Results obtained from the machine learning method is presented and compared with the results obtained with traditional methods.

arXiv comment6 pages, 7 figures, 5 tables

References in corpus (4)

• TensorFlow: Large-Scale Machine Learning on Heterogeneous Distributed Systems
• DELPHES 3, A modular framework for fast simulation of a generic collider experiment
• Search for narrow resonances and quantum black holes in inclusive and b-tagged dijet mass spectra from pp collisions at sqrt(s) = 7 TeV
• Data Unfolding Methods in High Energy Physics