Matching Convolutional Neural Networks without Priors about Data
This work addresses the challenge of applying CNNs to irregular graph data, which is incremental as it extends existing CNN concepts to new domains.
The authors tackled the problem of extending Convolutional Neural Networks to graph-structured data, achieving accuracy matching state-of-the-art CNNs on images without prior knowledge of 2D structure and obtaining a significant gain in accuracy on fMRI data compared to existing graph-based methods.
We propose an extension of Convolutional Neural Networks (CNNs) to graph-structured data, including strided convolutions and data augmentation on graphs. Our method matches the accuracy of state-of-the-art CNNs when applied on images, without any prior about their 2D regular structure. On fMRI data, we obtain a significant gain in accuracy compared with existing graph-based alternatives.