Machine learning

Anomaly detection — ML pipeline

A standalone Python pipeline ingests network flow CSVs, fits an Isolation Forest (or one-class SVM), and ships the per-flow inference scores as a static artifact. The Node bundle reads only that artifact — scikit-learn does not run in the deployed runtime. Below the artifact dashboard you can paste a streaming batch of flows and score them live against an in-memory 30-minute window using the same TS scorer as /api/detect.