Self-supervised Adversarial Purification for Graph Neural Networks

2026-05-22

AI Summary

Graph neural networks face a fundamental challenge in adversarial settings: the difficulty of simultaneously achieving high accuracy and robustness. Existing approaches often couple these objectives within a single classifier, limiting performance gains. This work proposes a self-supervised adversarial purification framework that decouples robustness learning from the classification task by introducing a dedicated GPR-GAE purification module to preprocess the input graph. The module innovatively integrates multi-scale generalized PageRank filters with a graph autoencoder and is trained in a self-supervised manner to enable structure-adaptive graph purification. Extensive experiments demonstrate that the proposed method significantly enhances robustness across multiple datasets and attack scenarios, validating the effectiveness of GPR-GAE as a plug-and-play purifier.

Abstract

Defending Graph Neural Networks (GNNs) against adversarial attacks requires balancing accuracy and robustness, a trade-off often mishandled by traditional methods like adversarial training that intertwine these conflicting objectives within a single classifier. To overcome this limitation, we propose a self-supervised adversarial purification framework. We separate robustness from the classifier by introducing a dedicated purifier, which cleanses the input data before classification. In contrast to prior adversarial purification methods, we propose GPR-GAE, a novel graph auto-encoder (GAE), as a specialized purifier trained with a self-supervised strategy, adapting to diverse graph structures in a data-driven manner. Utilizing multiple Generalized PageRank (GPR) filters, GPR-GAE captures diverse structural representations for robust and effective purification. Our multi-step purification process further facilitates GPR-GAE to achieve precise graph recovery and robust defense against structural perturbations. Experiments across diverse datasets and attack scenarios demonstrate the state-of-the-art robustness of GPR-GAE, showcasing it as an independent plug-and-play purifier for GNN classifiers. Our code can be found at https://github.com/woodavid31/GPR-GAE.

Problem

• Graph Neural Networks
• Adversarial Attacks
• Robustness
• Accuracy-Robustness Trade-off
• Adversarial Purification

Innovation

• self-supervised learning
• adversarial purification
• graph neural networks
• Generalized PageRank
• graph auto-encoder