Vertex finding in neutrino-nucleus interaction: A Model Architecture Comparison
Abstract
We compare different neural network architectures for Machine Learning (ML) algorithms designed to identify the neutrino interaction vertex position in the MINERvA detector. The architectures developed and optimized by hand are compared with the architectures developed in an automated way using the package "Multi-node Evolutionary Neural Networks for Deep Learning" (MENNDL), developed at Oak Ridge National Laboratory (ORNL). The two architectures resulted in a similar performance which suggests that the systematics associated with the optimized network architecture are small. Furthermore, we find that while the domain expert hand-tuned network was the best performer, the differences were negligible and the auto-generated networks performed well. There is always a trade-off between human, and computer resources for network optimization and this work suggests that automated optimization, assuming resources are available, provides a compelling way to save significant expert time.
Keywords
Cite
@article{arxiv.2201.02523,
title = {Vertex finding in neutrino-nucleus interaction: A Model Architecture Comparison},
author = {F. Akbar and A. Ghosh and S. Young and S. Akhter and Z. Ahmad Dar and V. Ansari and M. V. Ascencio and M. Sajjad Athar and A. Bodek and J. L. Bonilla and A. Bravar and H. Budd and G. Caceres and T. Cai and M. F. Carneiro and G. A. Díaz and J. Felix and L. Fields and A. Filkins and R. Fine and P. K. Gaura and R. Gran and D. A. Harris and D. Jena and S. Jena and J. Kleykamp and A. Klustová and D. Last and A. Lozano and X. G. Lu and E. Maher and S. Manly and W. A. Mann and K. S. McFarland and B. Messerly and J. Miller and O. Moreno and J. G. Morfín and J. K. Nelson and C. Nguyen and A. Olivier and V. Paolone and G. N. Perdue and K. J. Plows and M. A. Ramírez and D. Ruterbories and H. Su and V. S. Syrotenko and A. V. Waldron and B. Yaeggy and L. Zazueta},
journal= {arXiv preprint arXiv:2201.02523},
year = {2022}
}