Shared Data and Algorithms for Deep Learning in Fundamental Physics

Abstract

We introduce a Python package that provides simply and unified access to a collection of datasets from fundamental physics research - including particle physics, astroparticle physics, and hadron- and nuclear physics - for supervised machine learning studies. The datasets contain hadronic top quarks, cosmic-ray induced air showers, phase transitions in hadronic matter, and generator-level histories. While public datasets from multiple fundamental physics disciplines already exist, the common interface and provided reference models simplify future work on cross-disciplinary machine learning and transfer learning in fundamental physics. We discuss the design and structure and line out how additional datasets can be submitted for inclusion. As showcase application, we present a simple yet flexible graph-based neural network architecture that can easily be applied to a wide range of supervised learning tasks. We show that our approach reaches performance close to dedicated methods on all datasets. To simplify adaptation for various problems, we provide easy-to-follow instructions on how graph-based representations of data structures, relevant for fundamental physics, can be constructed and provide code implementations for several of them. Implementations are also provided for our proposed method and all reference algorithms.

14 pages, 3 figures, 5 tables - Version accepted by Computing and Software for Big Science

References in corpus (17)

• Very Deep Convolutional Networks for Large-Scale Image Recognition
• Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift
• Array Programming with NumPy
• Semi-Supervised Classification with Graph Convolutional Networks
• An Introduction to PYTHIA 8.2
• Delving Deep into Rectifiers: Surpassing Human-Level Performance on ImageNet Classification
• The Dark Machines Anomaly Score Challenge: Benchmark Data and Model Independent Event Classification for the Large Hadron Collider
• A Deep Learning-based Reconstruction of Cosmic Ray-induced Air Showers
• Lorentz Group Equivariant Neural Network for Particle Physics
• Lorentz Boost Networks: Autonomous Physics-Inspired Feature Engineering
• Deep learning stochastic processes with QCD phase transition
• A fast centrality-meter for heavy-ion collisions at the CBM experiment
• Accelerated Charged Particle Tracking with Graph Neural Networks on FPGAs
• Long Short-Term Memory (LSTM) networks with jet constituents for boosted top tagging at the LHC
• Playing Tag with ANN: Boosted Top Identification with Pattern Recognition
• Graph Generative Adversarial Networks for Sparse Data Generation in High Energy Physics
• Jet tagging in the Lund plane with graph networks

Cited by in corpus (8)

• Machine Learning and LHC Event Generation
• Ultra-High-Energy Cosmic Rays: The Intersection of the Cosmic and Energy Frontiers
• Exploring QCD matter in extreme conditions with Machine Learning
• Symmetries, Safety, and Self-Supervision
• OmniJet-$\alpha$: The first cross-task foundation model for particle physics
• Application of Graph Networks to background rejection in Imaging Air Cherenkov Telescopes
• Learning Tree Structures from Leaves For Particle Decay Reconstruction
• Deep learning techniques for Imaging Air Cherenkov Telescopes