Researchers have introduced AnomalyGFM, a novel graph foundation model designed for zero-shot and few-shot anomaly detection. This model addresses the limitations of existing generalist graph models by learning generalized knowledge to capture infrequent and irregular anomaly patterns across diverse graph datasets. AnomalyGFM achieves this by pre-training to align data-independent, learnable normal and abnormal class prototypes with node representation residuals, projecting node information into a unified feature space for consistent abnormality measurement. Experiments on 11 GAD datasets demonstrate its significant outperformance against state-of-the-art methods in both zero-shot and few-shot settings. AI
IMPACT Introduces a new foundation model for anomaly detection that improves generalization across diverse graph datasets, potentially impacting fields reliant on graph analysis.
RANK_REASON This is a research paper detailing a new model for anomaly detection. [lever_c_demoted from research: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
