Catching Every Ripple: Enhanced Anomaly Awareness via Dynamic Concept Adaptation

arXiv:2604.14726v2 Announce Type: replace-cross Abstract: Online anomaly detection (OAD) plays a pivotal role in real-time analytics and decision-making for evolving data streams. However, existing methods often rely on costly retraining and rigid decision boundaries, limiting their ability to...