Improving Fuzzy C-Means with Shadow Set

Fuzzy c-means (FCM) is an important algorithm in clustering. There has been much effort to improve FCM algorithm in efficiency. In this paper, we introduce the shadowed sets to FCM algorithm to enhance the efficiency and to improve the clustering results by properly eliminate some outliers during the process of clustering. The new fuzzy clustering algorithm, named shadow set-based FCM (S-FCM), presents better performance with respect to huge data sets especially to the data sets that do not exhibit definite number of clusters. Some experiments are carried to show the performance of the new algorithm.