Explainable Heterogeneous Anomaly Detection in Financial Networks via Adaptive Expert Routing
The paper introduces a system for financial anomaly detection that identifies not just that a crisis is happening, but *why*. It uses adaptive routing to categorize failures into four specific mechanisms (e.g., Price-Shock, Liquidity).Testing on U.S. equities, the model achieved high accuracy and provided an average lead time of 3.7 days for all six test events.