“…The vast majority of anomaly detection techniques used in maritime domain is data-driven. Considering the machine learning methods, the Bayesian network seems to be particularly useful in detection of single-point vessel anomalies (such as location, course, speed) [9,11,22,23], as well as to detect anomalies in the whole scenario of cooperation between a ship and pilot-boat [8] or piracy on oil platforms [4]. Neural networks are another popular classification-based technique, used to detect anomalies in ship's speed, position, and course [3,28].…”