Talos: A More Effective and Efficient Adversarial Defense for GNN Models Based on the Global Homophily of Graphs

TL;DR

Talos is a novel adversarial defense method for GNNs that leverages global graph homophily, outperforming existing defenses with minimal computational cost on large-scale graphs.

Contribution

The paper introduces Talos, a new defense technique that enhances global homophily in graphs, addressing limitations of local-focused and structure learning methods.

Findings

Talos significantly outperforms state-of-the-art defenses.

It maintains low computational overhead.

Effective on large-scale real-world graphs.

Abstract

Graph neural network (GNN) models play a pivotal role in numerous tasks involving graph-related data analysis. Despite their efficacy, similar to other deep learning models, GNNs are susceptible to adversarial attacks. Even minor perturbations in graph data can induce substantial alterations in model predictions. While existing research has explored various adversarial defense techniques for GNNs, the challenge of defending against adversarial attacks on real-world scale graph data remains largely unresolved. On one hand, methods reliant on graph purification and preprocessing tend to excessively emphasize local graph information, leading to sub-optimal defensive outcomes. On the other hand, approaches rooted in graph structure learning entail significant time overheads, rendering them impractical for large-scale graphs. In this paper, we propose a new defense method named Talos, which enhances the global, rather than local, homophily of graphs as a defense. Experiments show that the proposed approach notably outperforms state-of-the-art defense approaches, while imposing little computational overhead.