Spatial Graph Convolutional Networks

TL;DR

This paper introduces Spatial Graph Convolutional Networks (SGCNs), which incorporate spatial node features to improve learning from spatially-structured graphs, outperforming existing methods in image and chemical data tasks.

Contribution

The paper presents a GCN-inspired architecture that leverages node positions, generalizes GCNs and CNNs, and uses augmentation for better performance and invariance.

Findings

SGCN outperforms state-of-the-art methods on image classification.

SGCN improves chemical property prediction.

Incorporating spatial features enhances graph learning.

Abstract

Graph Convolutional Networks (GCNs) have recently become the primary choice for learning from graph-structured data, superseding hash fingerprints in representing chemical compounds. However, GCNs lack the ability to take into account the ordering of node neighbors, even when there is a geometric interpretation of the graph vertices 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 performance 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.