AHSG: Adversarial Attack on High-level Semantics in Graph Neural Networks

TL;DR

This paper introduces AHSG, a novel adversarial attack method on graph neural networks that preserves high-level semantics, making attacks more subtle and harder to detect while maintaining effectiveness.

Contribution

The paper proposes a new attack model that retains primary semantics in graphs, improving stealthiness and attack success compared to existing methods.

Findings

AHSG outperforms state-of-the-art attack methods in effectiveness.

The attack preserves primary semantics, making detection more difficult.

Experiments validate the attack's success against defended graph models.

Abstract

Adversarial attacks on Graph Neural Networks aim to perturb the performance of the learner by carefully modifying the graph topology and node attributes. Existing methods achieve attack stealthiness by constraining the modification budget and differences in graph properties. However, these methods typically disrupt task-relevant primary semantics directly, which results in low defensibility and detectability of the attack. In this paper, we propose an Adversarial Attack on High-level Semantics for Graph Neural Networks (AHSG), which is a graph structure attack model that ensures the retention of primary semantics. By combining latent representations with shared primary semantics, our model retains detectable attributes and relational patterns of the original graph while leveraging more subtle changes to carry out the attack. Then we use the Projected Gradient Descent algorithm to map the latent representations with attack effects to the adversarial graph. Through experiments on robust graph deep learning models equipped with defense strategies, we demonstrate that AHSG outperforms other state-of-the-art methods in attack effectiveness. Additionally, using Contextual Stochastic Block Models to detect the attacked graph further validates that our method preserves the primary semantics of the graph.