Adaptive Graph Convolutional Neural Networks

TLDR

Graph Convolutional Neural Networks extend CNNs to graph data, but existing filters assume fixed, shared graph structures, whereas real-world graphs vary in size and connectivity. The study introduces a generalized, flexible graph CNN that accepts arbitrary graph structures as input. During training, a task‑driven adaptive graph is learned for each input graph, using a distance‑metric learning approach to efficiently infer the graph structure. Experiments on nine datasets show the method improves convergence speed and predictive accuracy.

Abstract

Graph Convolutional Neural Networks (Graph CNNs) are generalizations of classical CNNs to handle graph data such as molecular data, point could and social networks. Current filters in graph CNNs are built for fixed and shared graph structure. However, for most real data, the graph structures varies in both size and connectivity. The paper proposes a generalized and flexible graph CNN taking data of arbitrary graph structure as input. In that way a task-driven adaptive graph is learned for each graph data while training. To efficiently learn the graph, a distance metric learning is proposed. Extensive experiments on nine graph-structured datasets have demonstrated the superior performance improvement on both convergence speed and predictive accuracy.