Active Learning

Anomalies correspond to the behavior of a system which does not conform to its expected or normal behavior. Identifying such anomalies from observed data, or the task of anomaly detection, is an important and often critical analysis task. This includes finding abnormalities in a medical image, fraudulent transactions in a credit card history, or structural defects in an aircraft’s engine. The importance of this problem has resulted in a large body of literature on this topic. However, given that the definition of an anomaly is strongly tied to the underlying application, the existing research is often embedded in the application domains, and it is unclear how methods developed for one domain would perform in another. The goal of this article is to provide a general introduction of the anomaly detection problem. We start with the basic formulation of the problem and then discuss the various extensions. In particular, we discuss the challenges associated with identifying anomalies in structured data and provide an overview of existing research in this area. We hope that this article will provide a better understanding of the different directions in which research has been done on this topic, and how techniques developed in one area can be applied in domains for which they were not intended to begin with.