§ art-014-equivariant-nn · April 17, 2026 · LITERATURE-REVIEW

Equivariant Neural Networks: Beyond Data Augmentation

Harikumar · 10 min read

Data augmentation handles symmetry by brute force. Equivariant neural networks handle it by design, baking group structure directly into weight-sharing patterns. A tour from G-CNNs to SE(3)-Transformers.

equivariance symmetry geometric deep learning

APR 15Group Theory Meets Machine Learning: An Introduction

7 min→

APR 20Representation Theory for Machine Learning Engineers

12 min→