Exploiting Learned Symmetries in Group Equivariant Convolutions
Computer Vision and Pattern Recognition
2021-06-10 v1
Abstract
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. All code is publicly available at github.com/Attila94/SepGrouPy.
Cite
@article{arxiv.2106.04914,
title = {Exploiting Learned Symmetries in Group Equivariant Convolutions},
author = {Attila Lengyel and Jan C. van Gemert},
journal= {arXiv preprint arXiv:2106.04914},
year = {2021}
}