Learning Broken Symmetries with Approximate Invariance
arXiv:2412.18773 · doi:10.1103/PhysRevD.111.072002
Abstract
Recognizing symmetries in data allows for significant boosts in neural network training, which is especially important where training data are limited. In many cases, however, the exact underlying symmetry is present only in an idealized dataset, and is broken in actual data, due to asymmetries in the detector, or varying response resolution as a function of particle momentum. Standard approaches, such as data augmentation or equivariant networks fail to represent the nature of the full, broken symmetry, effectively overconstraining the response of the neural network. We propose a learning model which balances the generality and asymptotic performance of unconstrained networks with the rapid learning of constrained networks. This is achieved through a dual-subnet structure, where one network is constrained by the symmetry and the other is not, along with a learned symmetry factor. In a simplified toy example that demonstrates violation of Lorentz invariance, our model learns as rapidly as symmetry-constrained networks but escapes its performance limitations.
7 pages, 8 figures
References in corpus (10)
- The automated computation of tree-level and next-to-leading order differential cross sections, and their matching to parton shower simulations
- Searching for Exotic Particles in High-Energy Physics with Deep Learning
- ConViT: Improving Vision Transformers with Soft Convolutional Inductive Biases
- Jet-Images -- Deep Learning Edition
- Energy Flow Networks: Deep Sets for Particle Jets
- Deep-learned Top Tagging with a Lorentz Layer
- An Efficient Lorentz Equivariant Graph Neural Network for Jet Tagging
- Search for resonant pair production of Higgs bosons in the final state using collisions at = 13 TeV with the ATLAS detector
- Does Lorentz-symmetric design boost network performance in jet physics?
- Semi-Equivariant GNN Architectures for Jet Tagging