We present a framework for deriving anomaly detection algorithms on timeseries data when the time and expression of anomalous behaviour is unknown. The framework is suited for problems in which individual machine learning ...