Adaptive outlierness for subspace outlier ranking.

Outlier mining is an important data analysis task to distinguish exceptional outliers from regular objects. However, in recent applications traditional outlier mining approaches miss outliers as they are hidden in subspace projections. In this work, we propose a novel outlier ranking based on the degree of deviation in subspaces. Object deviation is measured only in a selection of relevant subspaces and is based on adaptive neighborhoods in these subspaces. We show that our approach outperforms competing outlier ranking approaches by detecting outliers in arbitrary subspaces.