Scalable Graph Neural Networks via Bidirectional Propagation

TL;DR

This paper introduces GBP, a scalable graph neural network that employs bidirectional propagation, achieving sub-linear time complexity and state-of-the-art performance on large graphs with billions of edges.

Contribution

GBP is the first GNN method to combine bidirectional propagation with sub-linear complexity, enabling efficient learning on extremely large graphs.

Findings

GBP achieves state-of-the-art accuracy on large-scale graphs.

GBP trains and tests significantly faster than existing methods.

GBP handles graphs with over 60 million nodes and 1.8 billion edges in under 30 minutes.

Abstract

Graph Neural Networks (GNN) is an emerging field for learning on non-Euclidean data. Recently, there has been increased interest in designing GNN that scales to large graphs. Most existing methods use "graph sampling" or "layer-wise sampling" techniques to reduce training time. However, these methods still suffer from degrading performance and scalability problems when applying to graphs with billions of edges. This paper presents GBP, a scalable GNN that utilizes a localized bidirectional propagation process from both the feature vectors and the training/testing nodes. Theoretical analysis shows that GBP is the first method that achieves sub-linear time complexity for both the precomputation and the training phases. An extensive empirical study demonstrates that GBP achieves state-of-the-art performance with significantly less training/testing time. Most notably, GBP can deliver superior performance on a graph with over 60 million nodes and 1.8 billion edges in less than half an hour on a single machine. The codes of GBP can be found at https://github.com/chennnM/GBP .