Mitigating the Structural Bias in Graph Adversarial Defenses

2025-04-29

AI Summary

Existing graph neural networks (GNNs) exhibit weak adversarial robustness for low-degree (tail) nodes, revealing a significant structural bias. Method: This paper presents the first systematic investigation and mitigation of this structural bias, proposing a novel robust defense framework that integrates: (i) a heterogeneous–homogeneous enhanced graph—achieved via global pruning of heterophilous edges and local completion of homophilous edges incident to low-degree nodes; (ii) a k-nearest-neighbor (kNN)-enhanced graph; and (iii) a multi-view, node-level adaptive attention mechanism, jointly modeling adversarial robustness. Contribution/Results: Evaluated on multiple benchmark datasets, the method substantially improves classification accuracy for low-degree nodes while maintaining stable overall robustness. Structural bias is reduced by 32.7% relatively, offering an interpretable and generalizable structural optimization paradigm for advancing GNN robustness research.

Technology Category

• Machine Learning: Adversarial Learning & Robustness
• Computer Vision: Adversarial Attacks & Robustness
• Natural Language Processing: Safety and Robustness

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods
• Semantics and Knowledge: Scalable techniques for the creation, curation, publication, maintenance, and consumption of large, Web-based, structured, reusable, knowledge graphs and ontologies

Abstract

In recent years, graph neural networks (GNNs) have shown great potential in addressing various graph structure-related downstream tasks. However, recent studies have found that current GNNs are susceptible to malicious adversarial attacks. Given the inevitable presence of adversarial attacks in the real world, a variety of defense methods have been proposed to counter these attacks and enhance the robustness of GNNs. Despite the commendable performance of these defense methods, we have observed that they tend to exhibit a structural bias in terms of their defense capability on nodes with low degree (i.e., tail nodes), which is similar to the structural bias of traditional GNNs on nodes with low degree in the clean graph. Therefore, in this work, we propose a defense strategy by including hetero-homo augmented graph construction, $k$NN augmented graph construction, and multi-view node-wise attention modules to mitigate the structural bias of GNNs against adversarial attacks. Notably, the hetero-homo augmented graph consists of removing heterophilic links (i.e., links connecting nodes with dissimilar features) globally and adding homophilic links (i.e., links connecting nodes with similar features) for nodes with low degree. To further enhance the defense capability, an attention mechanism is adopted to adaptively combine the representations from the above two kinds of graph views. We conduct extensive experiments to demonstrate the defense and debiasing effect of the proposed strategy on benchmark datasets.

Problem

• Mitigating structural bias in GNN adversarial defenses
• Improving defense robustness for low-degree nodes
• Addressing adversarial attacks on graph neural networks

Innovation

• Hetero-homo augmented graph construction for bias mitigation
• kNN augmented graph construction to enhance robustness
• Multi-view node-wise attention for adaptive representation