TPC track denoising and recognition using convolutional neural networks

Author(s)

Gajdos, Matej ; da Luz, Hugo Natal ; Souza, Geovane G. A. ; Bregant, Marco

• Document type: Journal article
• Source: COMPUTER PHYSICS COMMUNICATIONS; v. 312, p. 9-pg., 2025-07-01.

Abstract

The capability of convolutional neural networks to remove spurious signals caused by electronic noise, microdischarges and other effects from experimental data obtained with Time Projection Chambers is studied. A generator of synthetic data for the training of the neural network is described and its performance is compared with the results obtained with a conventional algorithm. The Physical meaning of the data resulting from the neural network and conventional denoising algorithms is thoroughly analysed, demonstrating the potential of convolutional neural networks in the preparation of raw data for analysis. (AU)

• FAPESP's process: 20/04867-2 - High energy physics and instrumentation with the LHC-CERN
• Grantee: Marcelo Gameiro Munhoz
• Support Opportunities: Special Projects