What is one-class SVM and where is it used?

One-class SVM is used for anomaly detection by learning normal data patterns.

It trains only on normal data and identifies deviations as anomalies. It constructs a boundary around normal data in feature space using kernel methods.