The term „pattern-based anomaly detection“ is primarily found in the fields of Artificial Intelligence, Big Data and Smart Data, as well as cybercrime and cybersecurity. It involves automatically discovering unusual or suspicious events in large volumes of data.
Ordinarily, many processes, such as network connections in an office or production steps in a factory, follow similar patterns. Pattern-based anomaly detection uses artificial intelligence or software to first learn this „normality“. It then constantly examines new data and raises an alarm as soon as something unusual happens – for example, if a computer suddenly sends much larger amounts of data abroad than usual. This allows for the quick identification of a hacker attack or a technical error.
A clear example: In a factory, a system analyses a hundred machines and knows their usual power consumption patterns. If a machine suddenly starts up at night when it should be off, this is immediately recognised. This can prevent damage or attacks and optimise processes.
Pattern-based anomaly detection therefore helps companies to minimise risks and make their operations safer and more efficient.
Deutsch 
English (UK)