Adversarial Robustness of Link Sign Prediction in Signed Graphs

2024-01-19

AI Summary

This work uncovers an inherent vulnerability in signed graph neural networks (SGNNs) stemming from balance theory: its implicit assumption renders “balance-relevant information irreversible,” impeding existing defenses from accurately recovering edges perturbed by black-box attacks. To address this, we propose Balance-Attack—the first balance-theoretic black-box attack paradigm—and design BA-SGCL, a novel framework integrating balance-aware graph augmentation, NP-hard heuristic optimization, and contrastive learning to explicitly model and enhance structural robustness against balance violations. Evaluated on multiple real-world social network datasets, BA-SGCL achieves an average 8.3% improvement in link sign prediction accuracy and a 21.7% gain in edge recovery precision over state-of-the-art baselines. It is the first method to break the robustness bottleneck of signed graph representation learning under adversarial settings.

Technology Category

• Machine Learning: Adversarial Learning & Robustness
• Computer Vision: Adversarial Attacks & Robustness
• Data Mining & Knowledge Management: Graph Mining, Social Network Analysis & Community

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Signed graphs serve as fundamental data structures for representing positive and negative relationships in social networks, with signed graph neural networks (SGNNs) emerging as the primary tool for their analysis. Our investigation reveals that balance theory, while essential for modeling signed relationships in SGNNs, inadvertently introduces exploitable vulnerabilities to black-box attacks. To demonstrate this vulnerability, we propose balance-attack, a novel adversarial strategy specifically designed to compromise graph balance degree, and develop an efficient heuristic algorithm to solve the associated NP-hard optimization problem. While existing approaches attempt to restore attacked graphs through balance learning techniques, they face a critical challenge we term"Irreversibility of Balance-related Information,"where restored edges fail to align with original attack targets. To address this limitation, we introduce Balance Augmented-Signed Graph Contrastive Learning (BA-SGCL), an innovative framework that combines contrastive learning with balance augmentation techniques to achieve robust graph representations. By maintaining high balance degree in the latent space, BA-SGCL effectively circumvents the irreversibility challenge and enhances model resilience. Extensive experiments across multiple SGNN architectures and real-world datasets demonstrate both the effectiveness of our proposed balance-attack and the superior robustness of BA-SGCL, advancing the security and reliability of signed graph analysis in social networks. Datasets and codes of the proposed framework are at the github repository https://anonymous.4open.science/r/BA-SGCL-submit-DF41/.

Problem

• Vulnerabilities in signed graph neural networks from balance theory
• Irreversibility of balance-related information after adversarial attacks
• Need for robust representations against black-box adversarial strategies

Innovation

• Proposes balance-attack adversarial strategy exploiting vulnerabilities
• Develops efficient heuristic for NP-hard optimization problem
• Introduces BA-SGCL framework combining contrastive learning with augmentation