Exploiting Learned Symmetries in Group Equivariant Convolutions


Group Equivariant Convolutions (GConvs) enable convolutional neural networks to be equivariant to various transformation groups, but at an additional parameter and compute cost. We investigate the filter parameters learned by GConvs and find certain conditions under which they become highly redundant. We show that GConvs can be efficiently decomposed into depthwise separable convolutions while preserving equivariance properties and demonstrate improved performance and data efficiency on two datasets.

Proceedings of the 28th IEEE International Conference on Image Processing (ICIP 2021)
Jan van Gemert
Jan van Gemert
Associate Professor

Head of the CV Lab.