Informative Graph Structure Learning

TL;DR

This paper introduces InGSL, a novel graph structure learning method that balances similarity and diversity in edge construction, improving performance while reducing edge count and computational costs.

Contribution

InGSL is a plug-in module that enhances existing GSL methods by incorporating mutual information to optimize edge selection, addressing redundancy issues.

Findings

InGSL improves GSL performance on six benchmarks.

InGSL reduces edge count while maintaining or enhancing accuracy.

InGSL seamlessly integrates with existing GSL frameworks.

Abstract

The quality of graph-structured data is fundamental to the success of modern graph analysis techniques such as Graph Neural Networks (GNNs). However, real-world graph data is often suboptimal, suffering from issues such as noise and incomplete connections. Graph Structure Learning (GSL) has emerged as a promising technique that adaptively optimizes node connections. However, we observe that the effectiveness of GSL often comes at the cost of a dramatic expansion in edge count, resulting in significant storage and computational overhead. In this work, we reveal that this limitation stems from the prevalent use of similarity-based edge construction, which predominantly connects highly similar neighbors based on their embeddings, introducing substantial structure redundancy. To address this, we propose a novel Informative Graph Structure Learning method (InGSL), which jointly considers both similarity and diversity in edge construction by incorporating a mutual-information-guided learning strategy. Notably, InGSL serves as a plug-in module that can be seamlessly integrated into existing GSL frameworks. Through extensive experiments on six representative GSL methods, we demonstrate that InGSL achieves significant performance improvements at a reduced number of edges.