Equivariance-bridged SO(2)-Invariant Representation Learning using Graph Convolutional Network
This addresses the need for more robust and efficient rotation-invariant models in computer vision, reducing reliance on data augmentation, though it appears incremental by building on existing equivariance and GCN concepts.
The paper tackled the problem of achieving rotation invariance in CNNs without data augmentation by proposing a deep equivariance-bridged SO(2) invariant network using a Graph Convolutional Network (SWN-GCN) and Global Average Pooling, achieving state-of-the-art image classification performance on rotated MNIST and CIFAR-10 datasets with non-augmented training.
Training a Convolutional Neural Network (CNN) to be robust against rotation has mostly been done with data augmentation. In this paper, another progressive vision of research direction is highlighted to encourage less dependence on data augmentation by achieving structural rotational invariance of a network. The deep equivariance-bridged SO(2) invariant network is proposed to echo such vision. First, Self-Weighted Nearest Neighbors Graph Convolutional Network (SWN-GCN) is proposed to implement Graph Convolutional Network (GCN) on the graph representation of an image to acquire rotationally equivariant representation, as GCN is more suitable for constructing deeper network than spectral graph convolution-based approaches. Then, invariant representation is eventually obtained with Global Average Pooling (GAP), a permutation-invariant operation suitable for aggregating high-dimensional representations, over the equivariant set of vertices retrieved from SWN-GCN. Our method achieves the state-of-the-art image classification performance on rotated MNIST and CIFAR-10 images, where the models are trained with a non-augmented dataset only. Quantitative validations over invariance of the representations also demonstrate strong invariance of deep representations of SWN-GCN over rotations.