Training Robust Graph Neural Networks by Modeling Noise Dependencies

Chanyoung Park (Korea Advanced Institute of Science and Technology) · Sukwon Yun (University of North Carolina at Chapel Hill) · Yeonjun In (Korea Advanced Institute of Science & Technology) · Kanghoon Yoon (Korea Advanced Institute of Science & Technology) · Kibum Kim (Korea Advanced Institute of Science & Technology) · Sungchul Kim (Adobe Systems)

benchmark datasets causal relationshipsconventional noise modelsda-gnndata generating processdependency-aware noisegnns graph structurenode labelsnoise dependenciesnoise scenariosperformance degradationrobust gnnsvariational inference

In real-world applications, node features in graphs often contain noise from various sources, leading to significant performance degradation in GNNs. Although several methods have been developed to enhance robustness, they rely on the unrealistic assumption that noise in node features is independent of the graph structure and node labels, thereby limiting their applicability. To this end, we introduce a more realistic noise scenario, dependency-aware noise on graphs (DANG), where noise in node features create a chain of noise dependencies that propagates to the graph structure and node labels. We propose a novel robust GNN, DA-GNN, which captures the causal relationships among variables in the data generating process (DGP) of DANG using variational inference. In addition, we present new benchmark datasets that simulate DANG in real-world applications, enabling more practical research on robust GNNs. Extensive experiments demonstrate that DA-GNN consistently outperforms existing baselines across various noise scenarios, including both DANG and conventional noise models commonly considered in this field.