Group Equivariant Convolutional Networks
This work addresses the need for more efficient and expressive neural networks in computer vision by leveraging group symmetries, offering a novel approach rather than an incremental improvement.
The paper tackles the problem of sample complexity in convolutional neural networks by introducing Group equivariant Convolutional Neural Networks (G-CNNs), which exploit symmetries to reduce sample complexity and achieve state-of-the-art results on CIFAR10 and rotated MNIST.
We introduce Group equivariant Convolutional Neural Networks (G-CNNs), a natural generalization of convolutional neural networks that reduces sample complexity by exploiting symmetries. G-CNNs use G-convolutions, a new type of layer that enjoys a substantially higher degree of weight sharing than regular convolution layers. G-convolutions increase the expressive capacity of the network without increasing the number of parameters. Group convolution layers are easy to use and can be implemented with negligible computational overhead for discrete groups generated by translations, reflections and rotations. G-CNNs achieve state of the art results on CIFAR10 and rotated MNIST.