College of Sciences


Computer Science

Graduate Level


Publication Date





With the evolution in detector technologies and electronic components used in the Nuclear Physics field, experimental setups become larger and more complex. Faster electronics enable particle accelerator experiments to run with higher beam intensity, providing more interactions per time and more particles per interaction. However, the increased beam intensities present a challenge to particle detectors because of the higher amount of noise and uncorrelated signals. Higher noise levels lead to a more challenging particle reconstruction process by increasing the number of combinatorics to analyze and background signals to eliminate. On the other hand, increasing the beam intensity can provide physics outcomes faster if combined with a highly efficient track reconstruction process. Thus, a method that provides efficient tracking under high luminosity conditions can significantly reduce the amount of time required to conduct physics experiments.

In this poster, we present a machine learning (ML) approach for denoising data from particle tracking detectors to improve the track reconstruction efficiency of the CLAS12 detector at Jefferson Lab (JLab). A noise-reducing Convolutional Autoencoder was used to process data for standard experimental running conditions and showed significant improvements in track reconstruction efficiency (>15%). The studies were extended to synthetically generated data emulating much higher beam intensity and showed that the ML approach outperforms conventional algorithms, providing a significant increase in track reconstruction efficiency of up to 80%. This tremendous increase in reconstruction efficiency allows experiments to run at almost three times higher luminosity, leading to significant savings in time (about three times shorter) and money. The software developed by this work is now part of the CLASS12 workflow, assisting scientists of JLab and collaborating institutions.


Nuclear physics, Particle accelerators, Machine learning


Artificial Intelligence and Robotics | Nuclear



Download Poster (964 KB)

A Machine Learning Approach to Denoising Particle Detector Observations in Nuclear Physics