InfoGCL: Information-Aware Graph Contrastive Learning

TL;DR

InfoGCL introduces an information-aware framework for graph contrastive learning that unifies existing methods and enhances performance by applying the Information Bottleneck principle to preserve task-relevant information while reducing mutual information.

Contribution

It proposes a novel, unified framework based on the Information Bottleneck principle for graph contrastive learning, addressing customization challenges and improving task performance.

Findings

Outperforms state-of-the-art methods on node classification datasets.

Unifies various existing graph contrastive learning approaches.

Validates theoretical insights through empirical experiments.

Abstract

Various graph contrastive learning models have been proposed to improve the performance of learning tasks on graph datasets in recent years. While effective and prevalent, these models are usually carefully customized. In particular, although all recent researches create two contrastive views, they differ greatly in view augmentations, architectures, and objectives. It remains an open question how to build your graph contrastive learning model from scratch for particular graph learning tasks and datasets. In this work, we aim to fill this gap by studying how graph information is transformed and transferred during the contrastive learning process and proposing an information-aware graph contrastive learning framework called InfoGCL. The key point of this framework is to follow the Information Bottleneck principle to reduce the mutual information between contrastive parts while keeping task-relevant information intact at both the levels of the individual module and the entire framework so that the information loss during graph representation learning can be minimized. We show for the first time that all recent graph contrastive learning methods can be unified by our framework. We empirically validate our theoretical analysis on both node and graph classification benchmark datasets, and demonstrate that our algorithm significantly outperforms the state-of-the-arts.