Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
158 tokens/sec
GPT-4o
7 tokens/sec
Gemini 2.5 Pro Pro
45 tokens/sec
o3 Pro
4 tokens/sec
GPT-4.1 Pro
38 tokens/sec
DeepSeek R1 via Azure Pro
28 tokens/sec
2000 character limit reached

NuGraph2: A Graph Neural Network for Neutrino Physics Event Reconstruction (2403.11872v1)

Published 18 Mar 2024 in physics.data-an, cs.LG, and hep-ex

Abstract: Liquid Argon Time Projection Chamber (LArTPC) detector technology offers a wealth of high-resolution information on particle interactions, and leveraging that information to its full potential requires sophisticated automated reconstruction techniques. This article describes NuGraph2, a Graph Neural Network (GNN) for low-level reconstruction of simulated neutrino interactions in a LArTPC detector. Simulated neutrino interactions in the MicroBooNE detector geometry are described as heterogeneous graphs, with energy depositions on each detector plane forming nodes on planar subgraphs. The network utilizes a multi-head attention message-passing mechanism to perform background filtering and semantic labelling on these graph nodes, identifying those associated with the primary physics interaction with 98.0\% efficiency and labelling them according to particle type with 94.9\% efficiency. The network operates directly on detector observables across multiple 2D representations, but utilizes a 3D-context-aware mechanism to encourage consistency between these representations. Model inference takes 0.12 s/event on a CPU, and 0.005 s/event batched on a GPU. This architecture is designed to be a general-purpose solution for particle reconstruction in neutrino physics, with the potential for deployment across a broad range of detector technologies, and offers a core convolution engine that can be leveraged for a variety of tasks beyond the two described in this article.

Definition Search Book Streamline Icon: https://streamlinehq.com
References (26)
  1. B. Abi et al. (DUNE), Neutrino interaction classification with a convolutional neural network in the DUNE far detector, Phys. Rev. D 102, 092003 (2020a), arXiv:2006.15052 [physics.ins-det] .
  2. A. Abed Abud et al. (DUNE), Separation of track- and shower-like energy deposits in ProtoDUNE-SP using a convolutional neural network, Eur. Phys. J. C 82, 903 (2022), arXiv:2203.17053 [physics.ins-det] .
  3. C. Adams et al. (MicroBooNE), Deep neural network for pixel-level electromagnetic particle identification in the MicroBooNE liquid argon time projection chamber, Phys. Rev. D 99, 092001 (2019), arXiv:1808.07269 [hep-ex] .
  4. P. Abratenko et al. (MicroBooNE), Convolutional neural network for multiple particle identification in the MicroBooNE liquid argon time projection chamber, Phys. Rev. D 103, 092003 (2021a), arXiv:2010.08653 [hep-ex] .
  5. C. Choy, J. Gwak, and S. Savarese, 4d spatio-temporal convnets: Minkowski convolutional neural networks, in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (2019) pp. 3075–3084.
  6. P. Abratenko et al. (MicroBooNE), Semantic segmentation with a sparse convolutional neural network for event reconstruction in MicroBooNE, Phys. Rev. D 103, 052012 (2021b), arXiv:2012.08513 [physics.ins-det] .
  7. A. Abed Abud (DUNE), Sparse Convolutional Neural Networks for particle classification in ProtoDUNE-SP events, J. Phys. Conf. Ser. 2438, 012125 (2023).
  8. S. Farrell et al., Novel deep learning methods for track reconstruction, in 4th International Workshop Connecting The Dots 2018 (2018) arXiv:1810.06111 [hep-ex] .
  9. X. Ju et al. (Exa.TrkX), Performance of a geometric deep learning pipeline for HL-LHC particle tracking, Eur. Phys. J. C 81, 876 (2021), arXiv:2103.06995 [physics.data-an] .
  10. D. Murnane, S. Thais, and A. Thete, Equivariant Graph Neural Networks for Charged Particle Tracking, in 21th International Workshop on Advanced Computing and Analysis Techniques in Physics Research: AI meets Reality (2023) arXiv:2304.05293 [physics.ins-det] .
  11. K. Lieret and G. DeZoort, An Object Condensation Pipeline for Charged Particle Tracking at the High Luminosity LHC, in 26th International Conference on Computing in High Energy & Nuclear Physics (2023) arXiv:2309.16754 [physics.data-an] .
  12. V. Hewes et al. (Exa.TrkX), Graph neural network for object reconstruction in liquid argon time projection chambers, EPJ Web of Conferences 251, 03054 (2021).
  13. G. Cerati (MicroBooNE), MicroBooNE Public Data Sets: a Collaborative Tool for LArTPC Software Development, in 26th International Conference on Computing in High Energy & Nuclear Physics (2023) arXiv:2309.15362 [hep-ex] .
  14. R. Acciarri et al. (MicroBooNE), Design and Construction of the MicroBooNE Detector, JINST 12 (02), P02017, arXiv:1612.05824 [physics.ins-det] .
  15. R. Acciarri et al. (MicroBooNE), The Pandora multi-algorithm approach to automated pattern recognition of cosmic-ray muon and neutrino events in the MicroBooNE detector, Eur. Phys. J. C 78, 82 (2018), arXiv:1708.03135 [hep-ex] .
  16. https://github.com/vhewes/numl.
  17. https://github.com/vhewes/pynuml.
  18. D. Misra, Mish: A self regularized non-monotonic activation function (2020), arXiv:1908.08681 [cs.LG] .
  19. A. Krizhevsky, I. Sutskever, and G. E. Hinton, Imagenet classification with deep convolutional neural networks, Commun. ACM 60, 84–90 (2017).
  20. M. Fey and J. E. Lenssen, Fast graph representation learning with pytorch geometric (2019), arXiv:1903.02428 [cs.LG] .
  21. I. Loshchilov and F. Hutter, Decoupled weight decay regularization (2019), arXiv:1711.05101 [cs.LG] .
  22. L. N. Smith and N. Topin, Super-convergence: Very fast training of neural networks using large learning rates (2018), arXiv:1708.07120 [cs.LG] .
  23. A. Kendall, Y. Gal, and R. Cipolla, Multi-task learning using uncertainty to weigh losses for scene geometry and semantics (2018), arXiv:1705.07115 [cs.CV] .
  24. E. L. Snider and G. Petrillo, LArSoft: Toolkit for Simulation, Reconstruction and Analysis of Liquid Argon TPC Neutrino Detectors, J. Phys. Conf. Ser. 898, 042057 (2017).
  25. P. Abratenko et al. (MicroBooNE), MicroBooNE BNB Inclusive Overlay Sample (No Wire Info), 10.5281/zenodo.8370883 (2023).
  26. P. Abratenko et al. (MicroBooNE), MicroBooNE BNB Electron Neutrino Overlay Sample (No Wire Info), 10.5281/zenodo.7261921 (2022).

Summary

We haven't generated a summary for this paper yet.