Detecting outlying subspaces for high-dimensional data: a heuristic search approach

摘要: [Abstract]: In this paper, we identify a new task for studying the out-lying degree of high-dimensional data, i.e. finding sub-spaces (subset features) in which given points are out-liers, and propose novel detection algorithm, called High-D Outlying subspace Detection (HighDOD). We measure outlying point using sum distances between its k nearest neighbors. Heuristic pruning strategies proposed to realize fast search an efficient dynamic search method with sample-based learning process has been im- plemented. Experimental results show that HighDOD is outperforms other searching alternatives such as naive top-down, bottom-up random methods. Points these sparse subspaces assumed be the outliers. While knowing data the outliers can be useful, many applications, it more important given point outlier, motivates proposal a new technique paper handle task.