English

The High Cost of Data Augmentation for Learning Equivariant Models

Numerical Analysis 2026-02-04 v1 Numerical Analysis

Abstract

According to Noether's theorem the presence of a continuous symmetry in a Hamiltonian systems is equivalent to the existence of a conserved quantity, yet these symmetries are not always explicitly enforced in data-driven models. There remains a debate whether or not encoding of symmetry into a model architecture is the optimal approach. A competing approach is to target approximate symmetry through data augmentation. In this work, we study two approaches aimed at improving the symmetry properties of such an approximation scheme: one based on a quadrature rule for the Haar measure on the compact Lie group encoding the continuous symmetry of interest and one based on a random sampling of that Haar measure. We demonstrate both theoretically and empirically that the quadrature augmentation leads to exact symmetry preservation in polynomial models, while the random augmentation has only square-root convergence of the symmetrization error.

Cite

@article{arxiv.2602.03118,
  title  = {The High Cost of Data Augmentation for Learning Equivariant Models},
  author = {Henri Klintebäck and Christoph Ortner and Lior Silberman},
  journal= {arXiv preprint arXiv:2602.03118},
  year   = {2026}
}

Comments

33 pages, 13 figures

R2 v1 2026-07-01T09:33:30.859Z