Spatial Graph Convolutional Networks
share
Graph Convolutional Networks (GCNs) have recently be- come the primary choice for learning from graph-structured data, super- seding hash fingerprints in representing chemical compounds. However, GCNs lack the ability to take into account the ordering of node neigh- bors, even when there is a geometric interpretation of the graph ver- tices that provides an order based on their spatial positions. To remedy this issue, we propose Spatial Graph Convolutional Network (SGCN) which uses spatial features to efficiently learn from graphs that can be naturally located in space. Our contribution is threefold: we propose a GCN-inspired architecture which (i) leverages node positions, (ii) is a proper generalization of both GCNs and Convolutional Neural Networks (CNNs), (iii) benefits from augmentation which further improves the per- formance and assures invariance with respect to the desired properties. Empirically, SGCN outperforms state-of-the-art graph-based methods on image classification and chemical tasks.
READ FULL TEXT
