Search papers, labs, and topics across Lattice.
This paper introduces OTTA-DGAD, a novel approach for online test-time adaptation in dynamic graph anomaly detection that addresses the limitations of existing methods by leveraging evolving representations of normal and anomalous patterns. By utilizing a memory buffer to store both general patterns from source domains and new patterns from sequentially arriving target data, OTTA-DGAD effectively identifies anomalies without the need for costly retraining. The method achieves state-of-the-art performance across ten real-world datasets, showcasing its ability to adapt to domain-specific changes in an online setting.
Anomaly detection can now adapt in real-time to evolving patterns without retraining, achieving state-of-the-art results on diverse datasets.
Generalizable dynamic graph anomaly detection (DGAD) enables pretrained detectors to identify anomalies in unseen target domains without costly retraining. However, existing methods often fail for two reasons. First, they mainly rely on domain-agnostic patterns and miss domain-specific patterns that keep evolving. Second, they assume access to the full target domain data, whereas in more practical online test-time adaptation settings, target data arrive sequentially in unlabeled chunks. To address these limitations, we formulate online test-time adaptation for generalizable DGAD and propose OTTA-DGAD. OTTA-DGAD first extracts dynamic prototypes, i.e., evolving representations of normal and anomalous patterns, from temporal ego-graphs and stores them in a memory buffer. The buffer selectively retains general patterns shared across the source domains used for pretraining while incorporating new patterns from the target domain. An anomaly scorer then compares incoming edge representations against these prototypes to identify both general and domain-specific anomalies. During adaptation, OTTA-DGAD updates the memory buffer using reliable pseudo-labels identified through confidence-based detection. It further enriches each target chunk with relevant representations retained from previous chunks, compensating for information loss resulting from the sequential arrival of data. Extensive experiments under strict test-then-adapt OTTA settings demonstrate state-of-the-art performance on ten real-world datasets from diverse domains.
