Ανίχνευση ασυνήθιστης συμπεριφοράς στην εξόρυξη δεδομένων
Anomaly detection in data mining
The term outlier or anomaly detection is used to describe techniques that detect objects, events or observations that do not follow the general rules of the data we observe and are characterized as anomalies. Anomalies arise due to mechanical faults, changes in system behavior, fraudulent behavior, human error, instrument error or simply through natural deviations in populations. Their detection can identify system faults and fraud before they escalate with potentially catastrophic consequences. Many anomaly detection techniques have been specifically developed for certain application domains, while others are more generic. This thesis tries to provide a structured and comprehensive overview of the research on anomaly detection. For each category of anomaly detection techniques, the basic technique and the assumptions used to differentiate between normal and anomaly behavior are presented. In addition, for each category various methods of anomaly detection, which are based on the corresponding basic technique are presented, and their advantages and disadvantages are determined. Finally, various methods are applied, and their performance in detecting anomalies is compared.