FAME: Failure-Aware Mixture-of-Experts for Message-Level Log Anomaly Detection
Researchers have developed FAME, a novel framework for message-level log anomaly detection that significantly reduces the need for manual labeling. This system utilizes a Mixture-of-Experts approach, employing large language models offline to partition log templates into failure domains. FAME trains lightweight routers and domain experts that can be run on-premise, achieving high F1 scores on benchmark datasets like BGL and Thunderbird while drastically cutting down annotation effort. AI
IMPACT Enables more efficient and precise anomaly detection in production systems by reducing reliance on extensive manual labeling.