Exploring Domain Generalization and Subpopulation Shift for Generalizable Graph-Level Anomaly Detection

Graph-level anomaly detection (GLAD), which identifies rare or atypical graphs within a graph set, is crucial for applications such as image analysis, industrial defect inspection and fraud detection. However, existing GLAD approaches typically rely on the in-distribution hypothesis while lacking generalization capability for out-of-distribution (OOD) scenarios (e.g., different graph sizes), which largely limits the application in the real world. For the first time, we formulate the OOD generalization problem for GLAD, where testing graph data exhibit significant distributional shifts from training data. To tackle two common types of distributional shifts, domain generalization and subpopulation shift, we propose the $\textbf{F}$ine-$\textbf{G}$rained $\textbf{S}$ubpopulation $\textbf{G}$raph-$\textbf{L}$evel $\textbf{A}$nomaly $\textbf{D}$etection ($\textbf{FGS-GLAD}$). First, we propose a $\textbf{G}$raph $\textbf{I}$nformation $\textbf{B}$ottleneck-based $\textbf{A}$nomaly $\textbf{D}$etection Module ($\textbf{GIB4AD}$) that implements graph reverse distillation and graph information bottleneck on the graph to enhance task-relevant feature extraction for domain generalization. Second, We propose a $\textbf{F}$ine-$\textbf{G}$rained $\textbf{S}$ubpoulation $\textbf{I}$nference Module ($\textbf{FGSI}$) to predict fine-grained subpopulations and focus on critical inter-subpopulation features through a supervised contrastive mechanism. Experiments on seven benchmark datasets and ten baselines demonstrate our model's superiority in handling domain generalization and subpopulation shift, advancing graph-level anomaly detection for real-world applications.