Learning Invariant Rules from Data for Interpretable Anomaly Detection