Example-Based DB-Outlier Detection from High Dimensional Datasets

Outlier detection is an important problem that has applications in many fields. High dimensional datasets are common in such applications. Among the existing outlier detection methods, Distance-Based outlier (DB-Outlier) detection is one of the most generalizable and simplest approaches. It finds outliers by calculating distances between data points. However, in high dimensional space, data distribution is sparse, so every data point becomes a good outlier candidate. It has been shown that meaningful outliers are likely to be identified by examining the behavior of the data in low dimensional projections. On the other hand, Example-Based outlier detection method is promising in discovering the hidden user view of outliers. In this paper, we present a new method to detect DB-Outliers in high dimensional datasets based on user examples. The method finds a subspace where user examples are outstanding more significantly than in any other subspaces, and reports outliers detected in this subspace.