Robust clustering algorithm for suppression of outliers
Benson S. Y. Lam, Hong Yan · 2005
The fuzzy c-means clustering algorithm has been widely used in many data classification problems. However, the performance of this algorithm is easily degraded if an outlier is present. To solve this problem, we introduce the modified l/sub 2/ norm in the clustering formulation, which can overcome the influence of outliers effectively. Our experiments show that the proposed method is able to generate accurate results even if outliers are present.