Improving Prediction of Distance-Based Outliers

An unsupervised distance-based outlier detection method that finds the top n outliers of a large and high-dimensional data set D, is presented. The method provides a subset R of the data set, called robust solving set, that contains the top n outliers and can be used to predict if a new unseen object p is an outlier or not by computing the distances of p to only the objects in R. Experimental results show that the prediction accuracy of the robust solving set is comparable with that obtained by using the overall data set