Abstract:

We discuss the use of approximation and of ensemble techniques for unsupervised outlier detection. To introduce the field, we will briefly sketch the data mining task of unsupervised outlier detection and discuss some basic considerations about ensemble techniques. We give an overview on existing approaches to using ensemble techniques for outlier detection as well as on the challenges in doing so. We focus then on three recent ensemble methods for outlier detection, featuring different meta-approaches to get diverse ensemble members based on approximations.