Description

Outlier detection has been an active area of research for few decades. We propose a new definition of outlier useful for high-dimensional data. Given a dictionary of atoms learned using the sparse coding objective, the outlierness of a data point depends jointly on two factors: the frequency of each atom in reconstructing all data points (or its negative log activity ratio, NLAR) and the strength by which it is used in reconstructing the current point. A Rarity based Outlier Detection algorithm in a Sparse coding framework (RODS) is developed that consists of two components, NLAR learning and outlier scoring. The algorithm is unsupervised both the offline and online variants are presented. It is governed by very few parameters and operates in linear time. We demonstrate the superior performance of the RODS in comparison with various state-of-the-art outlier detection algorithms on a number of benchmark datasets. We also demonstrate its effectiveness using three real-world case studies: saliency detection in images, abnormal event detection in videos and change detection in data streams. Evaluations show the RODS out performs competing algorithms reported in the outlier detection, saliency detection, video event detection and change detection literature.