Local Subspace Based Outlier Detection

Abstract. Existing studies in outlier detection mostly focus on detecting outliers in full feature space. But most algorithms tend to break down in highdimensional feature spaces because classes of objects often exist in specific subspace of the original feature space. Therefore, subspace outlier detection has been recently defined. As a novel solution to tackle this problem, we propose here a local subspace based outlier detection technique, which uses different subspaces for different objects. Using this concept we adopt local density based outlier detection to cope with high-dimensional data. A broad experimental evaluation shows that this approach yields results of significantly better quality than existing algorithms.

• Type: Journal
• Year: 2009
• Where: IC3
• Authors: Ankur Agrawal
• Keywords: Applied Computing, Feature Space, IC3 2009, Outlier Detection, Outlier Detection Technique