Learning (Approximately) Equivariant Networks via Constrained Optimization

NeurIPSOral2025

Authors
Andrei Manolache, Luiz F. O. Chamon, Mathias Niepert
Affiliation
Universität Stuttgart
Venue
NeurIPS 2025
Track
Oral

TL;DR

We introduce Adaptive Constrained Equivariance, a homotopy-inspired constrained optimization apprach for training equivariant neural networks.

Opening excerpt from the authors’ abstract. source

Read the paper

Topics

optimization

← All NeurIPS 2025 Oral papers · Browse the whole archive