Adversarial Graph Augmentation to Improve Graph Contrastive Learning

TL;DR

This paper introduces adversarial graph augmentation for contrastive learning of GNNs, improving robustness and transferability by avoiding redundant features, validated across multiple datasets and tasks.

Contribution

It proposes a novel adversarial augmentation strategy for GCL that enhances GNN performance and robustness, with theoretical backing and practical implementation.

Findings

Achieves up to 14% performance improvement in unsupervised learning.

Demonstrates robustness across 18 benchmark datasets.

Improves transfer and semi-supervised learning results.

Abstract

Self-supervised learning of graph neural networks (GNN) is in great need because of the widespread label scarcity issue in real-world graph/network data. Graph contrastive learning (GCL), by training GNNs to maximize the correspondence between the representations of the same graph in its different augmented forms, may yield robust and transferable GNNs even without using labels. However, GNNs trained by traditional GCL often risk capturing redundant graph features and thus may be brittle and provide sub-par performance in downstream tasks. Here, we propose a novel principle, termed adversarial-GCL (AD-GCL), which enables GNNs to avoid capturing redundant information during the training by optimizing adversarial graph augmentation strategies used in GCL. We pair AD-GCL with theoretical explanations and design a practical instantiation based on trainable edge-dropping graph augmentation. We experimentally validate AD-GCL by comparing with the state-of-the-art GCL methods and achieve performance gains of up-to $14\%$ in unsupervised, $6\%$ in transfer, and $3\%$ in semi-supervised learning settings overall with 18 different benchmark datasets for the tasks of molecule property regression and classification, and social network classification.