TDGIA:Effective Injection Attacks on Graph Neural Networks

TL;DR

This paper introduces TDGIA, a novel injection attack method that significantly compromises GNNs by injecting adversarial nodes, revealing topological vulnerabilities and outperforming existing attacks on large-scale datasets.

Contribution

The paper proposes TDGIA, a new effective injection attack on GNNs that exploits topological vulnerabilities and outperforms existing methods in attacking defense models.

Findings

TDGIA significantly outperforms baseline attacks.

Injected nodes cause over double the performance drop.

Effective attack across multiple GNN models.

Abstract

Graph Neural Networks (GNNs) have achieved promising performance in various real-world applications. However, recent studies have shown that GNNs are vulnerable to adversarial attacks. In this paper, we study a recently-introduced realistic attack scenario on graphs -- graph injection attack (GIA). In the GIA scenario, the adversary is not able to modify the existing link structure and node attributes of the input graph, instead the attack is performed by injecting adversarial nodes into it. We present an analysis on the topological vulnerability of GNNs under GIA setting, based on which we propose the Topological Defective Graph Injection Attack (TDGIA) for effective injection attacks. TDGIA first introduces the topological defective edge selection strategy to choose the original nodes for connecting with the injected ones. It then designs the smooth feature optimization objective to generate the features for the injected nodes. Extensive experiments on large-scale datasets show that TDGIA can consistently and significantly outperform various attack baselines in attacking dozens of defense GNN models. Notably, the performance drop on target GNNs resultant from TDGIA is more than double the damage brought by the best attack solution among hundreds of submissions on KDD-CUP 2020.