Top-k Distance-Based Outlier Detection on Uncertain Data

This paper studies the problem of top-k distance-based outlier detection on uncertain data. In this work, an uncertain object is modelled by a Gaussian probability density function. Since the Naive approach is very expensive due to costly distance function between uncertain objects, a populated-cell list (PC-list) based top-k distance-based outlier detection approach is proposed in this work. Where PC-list is a sorted list of non-empty cells of a grid (grid is used to index dataset objects). Using PC-list, the top-k outlier detection algorithm needs to consider only a fraction of dataset objects and hence quickly identifies candidate objects for top-k outliers. An extensive empirical study shows that our proposed approach is effective, efficient and scalable.