Search papers, labs, and topics across Lattice.
This paper introduces ProMoS, an unsupervised generalist graph anomaly detection (GAD) framework that transfers across new graphs without requiring annotations or few-shot support. ProMoS uses knowledge distillation to transfer normality priors from a frozen self-supervised GNN teacher to a mixture-of-students model. The method aligns teacher and student models in a shared prototype space using prototype-guided soft-label distillation, improving cross-graph generalization and enabling zero-shot anomaly detection on unseen graphs.
Unsupervised graph anomaly detection is now possible across diverse graphs, achieving zero-shot generalization without relying on labeled data or few-shot examples.
Driven by the pressing demand for graph anomaly detection (GAD) in high-stakes domains, the generalist GAD paradigm, which trains a single detector transferable across new graphs, has recently gained growing attention. However, existing methods often rely on scarce and costly annotations for training and sometimes even require few-shot support at inference, which limits their robustness to diverse and unseen anomaly patterns. To address this limitation, we introduce ProMoS, the first unsupervised generalist GAD framework, which detects anomalies by modeling the abundant normality in unlabeled data. ProMoS adopts a knowledge-distillation paradigm to distill normality priors from a frozen self-supervised graph neural network (GNN) teacher to a mixture-of-students model with shared global and lightweight personalized branches, enabling efficient and expressive normality modeling without learning from scratch. We further propose prototype-guided soft-label distillation to align teacher and student in a shared prototype space, enhancing cross-graph generalizability. During inference, ProMoS performs zero-shot anomaly detection on unseen graphs via distillation bias and prototype geometric deviation. Extensive experiments show the effectiveness and efficiency of ProMoS, charting a practical path toward label-free, zero-shot generalist GAD.
