Rough-Based Semi-supervised Outlier Detection

With the help of some labeled samples and rough C-means clustering, a rough-based semi-supervised outlier detection (RBSSOD) is proposed, which integrates the advantage of semi-supervised outlier detection (SSOD) and rough C-means clustering. This method takes into account the information of labeled points, as well as the points located in boundary area of each cluster, which can be further discussed the possibility to be reassigned as outliers. Experiment results show that our method not only keep, or improve precision and false alarm rate but also speed up the learning process.