GJDNet: Robust Graph Neural Networks via Joint Disentangled Learning Against Adversarial Attacks
2026-06-01 • Machine Learning
Machine LearningArtificial Intelligence
AI summaryⓘ
The authors study how Graph Neural Networks (GNNs) can be tricked by attacks that mess up the usual connection patterns in networks. They found existing defenses don't work well because they don't consider how these attacks change the relationships between node features and structure. To fix this, the authors created a new method, GJDNet, that separates out the true node information from the misleading parts and makes the classification rules clearer and more stable. Their approach improves GNN robustness on various kinds of network structures, supported by theory and experiments.
Graph Neural NetworksAdversarial AttacksAssortativityNode ClassificationRepresentation LearningNeighborhood AggregationStructural PerturbationDisentangled RepresentationDecision BoundariesRobustness
Authors
Canyixing Cui, Tao Wu, Xingping Xian, Xiao-Ke Xu, Mao Wang, Weina Niu
Abstract
Graph Neural Networks (GNNs) are vulnerable to adversarial attacks, which inherently invert connectivity patterns by introducing disassortative edges in assortative graphs and assortative edges in disassortative graphs. This structural inversion creates structure-feature mismatches that disrupt neighborhood aggregation across different graph types. However, we find that existing defenses are limited, as they either treat neighborhoods as monolithic under fixed assortativity assumptions or rely on standard softmax classifiers that fail to account for perturbation-induced representation shifts. To further exploit this observation, we adopt a robustness perspective that jointly disentangles node representations and decision spaces, isolating perturbation effects while enforcing well-separated decision regions. Based on this principle, we propose Graph Joint Disentanglement Network (GJDNet), a unified framework for robust node classification across diverse graph assortativity regimes. GJDNet enhances robustness at both representation and decision levels: it employs feature-driven soft structural disentanglement with skewness-aware neighbor filtering to suppress perturbation-induced structure-feature mismatches, and introduces a Spherical Decision Boundary (SDB) to promote intra-class compactness and inter-class separation in the embedding space, thereby stabilizing decision boundaries under perturbations. Theoretical analysis provides insights into the effectiveness of the proposed disentangled representation and decision mechanisms, while extensive experiments demonstrate that GJDNet consistently achieves strong robustness across graphs with different connectivity regimes.