Matching Convolutional Neural Networks without Priors about Data

02/27/2018

∙

by   Carlos Eduardo Rosar Kos Lassance, et al.

∙

0

∙

share

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.