How unconstrained machine-learning models learn physical symmetries

M. Domina, J. W. Abbott, P. Pegolo, F. Bigi, M. Ceriotti · 2026

Read How unconstrained machine-learning models learn physical symmetries by M. Domina, J. W. Abbott, P. Pegolo, F. Bigi, M. Ceriotti (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-24638.txt.

Description

Domina and co-authors measure how unconstrained learning models learn physical symmetries. Many physics models enforce symmetries exactly, yet unconstrained models often match them. With simple data augmentation, such models learn near-equivariant behaviour. The authors give rigorous metrics for the symmetry content of learned representations. They apply the metrics to two transformer models on point clouds. The framework also diagnoses spectral failure modes during training.

Catalog record

CREATORM. Domina, J. W. Abbott, P. Pegolo, F. Bigi, M. Ceriotti
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE55244
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2603.24638

Browse the full library · About the Tower of Babel