Break the Ceiling: Stronger Multi-scale Deep Graph Convolutional Networks

TL;DR

This paper introduces two novel deep multi-scale graph convolutional architectures that theoretically and empirically enhance the expressive power of GCNs, leading to improved node classification performance.

Contribution

It provides a theoretical analysis of GCN limitations and proposes scalable, multi-scale deep architectures with proven effectiveness on benchmark tasks.

Findings

Achieved better node classification accuracy than state-of-the-art methods.

Demonstrated the scalability of the proposed architectures to deeper networks.

Established the theoretical equivalence of the two architectures under certain conditions.

Abstract

Recently, neural network based approaches have achieved significant improvement for solving large, complex, graph-structured problems. However, their bottlenecks still need to be addressed, and the advantages of multi-scale information and deep architectures have not been sufficiently exploited. In this paper, we theoretically analyze how existing Graph Convolutional Networks (GCNs) have limited expressive power due to the constraint of the activation functions and their architectures. We generalize spectral graph convolution and deep GCN in block Krylov subspace forms and devise two architectures, both with the potential to be scaled deeper but each making use of the multi-scale information in different ways. We further show that the equivalence of these two architectures can be established under certain conditions. On several node classification tasks, with or without the help of validation, the two new architectures achieve better performance compared to many state-of-the-art methods.