Anomaly Detection

#Terms

WHAT IS… ?

Anomaly detection is the identification of rare items, events or observations which raise suspicions by differing significantly from the majority of the data; it is a vital component in many fields. It can be used in computer security to identify potential intruders, medical imaging, bioinformatics, financial forecasting, fraud detection and business intelligence applications.

HOW IS IT USED ?

Anomaly detection is critical for many real-time event processing applications. The typical approach to this problem is to use the existing techniques for detecting anomalies in time series (e.g., linear and nonlinear autoregression, ARIMA models, moving windows) or to use techniques designed for one-class classification problems. This technique can be augmented using pattern discovery and data mining tools such as K-means clustering and decision trees. It can also be used in intrusion detection systems (IDS) used in computer security to identify potential intruders, in medical imaging to detect tumours; in bioinformatics to identify aberrations such as microsatellites and SNPs, and also in financial forecasting, fraud detection and business intelligence application.