Filtering of pulses from particle detectors using neural networks by dimensionality reduction

Abstract

This article presents a comparison between different filtering methods based on dimensionality reduction for pulses generated on particle detectors . This reduction has been carried out using Neural Networks (NNs). In particular, three topologies have been used: Autoencoders (AEs), Denoising Autoencoders (DAEs) and Restricted Boltzmann Machines (RBMs). A detailed explanation of these methods, a noise reduction analysis, filtering with simulated data and processing of pulses from a neutron detector have been carried out to verify the feasibility of using these NNs as pulse filters.