Top K Enhanced Reinforcement Learning Attacks on Heterogeneous Graph Node Classification

TL;DR

This paper introduces HeteroKRLAttack, a reinforcement learning-based black-box attack method that uses a Top-K algorithm to effectively disrupt node classification in heterogeneous graphs, revealing vulnerabilities in current GNN models.

Contribution

The paper presents a novel reinforcement learning attack framework with a Top-K algorithm tailored for heterogeneous graphs, improving attack efficiency and effectiveness.

Findings

HeteroKRLAttack significantly reduces classification accuracy on multiple datasets.

The Top-K algorithm enhances attack performance compared to baseline methods.

The study highlights vulnerabilities in current GNN models on heterogeneous graphs.

Abstract

Graph Neural Networks (GNNs) have attracted substantial interest due to their exceptional performance on graph-based data. However, their robustness, especially on heterogeneous graphs, remains underexplored, particularly against adversarial attacks. This paper proposes HeteroKRLAttack, a targeted evasion black-box attack method for heterogeneous graphs. By integrating reinforcement learning with a Top-K algorithm to reduce the action space, our method efficiently identifies effective attack strategies to disrupt node classification tasks. We validate the effectiveness of HeteroKRLAttack through experiments on multiple heterogeneous graph datasets, showing significant reductions in classification accuracy compared to baseline methods. An ablation study underscores the critical role of the Top-K algorithm in enhancing attack performance. Our findings highlight potential vulnerabilities in current models and provide guidance for future defense strategies against adversarial attacks on heterogeneous graphs.