Determining the Multiplicities of Muons in DECOR Events by Means of Deep Machine Learning

Abstract

The DECOR coordinate-tracking detector is designed to register charged cosmic ray particles in wide zenith angles. Measurements by the installation are currently analyzed manually, affecting its performance. The use of deep machine learning allows automated processing and larger samples of processed data. The artificial neural network (ANN) architectures considered in this work have displayed high accuracy in counting the multiplicity of muons in data from the DECOR facility. Estimates are given of ANN performance for events with different muon multiplicities. The accuracy is 1 track for 5–6 particles and 7 tracks for more than 100 particles.