Exploiting Learned Symmetries in Group Equivariant Convolutions

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.

On Translation Invariance in CNNs: Convolutional Layers Can Exploit Absolute Spatial Location

We show we can prevent CNNs from exploiting absolute location through image boundary effects.