Aggregation of Standard and Entropy Based Fuzzy c-Means Clustering by a Modified Objective Function

A generalized fuzzy c-means (FCM) clustering is proposed by modifying the standard FCM objective function and introducing some simplifications. FCM clustering results in very fuzzy partitions for data points that are far from all cluster centroids. This property distinguishes FCM from Gaussian mixture models or entropy based clustering. The generalized FCM clustering aims at aggregating standard FCM and entropy based FCM so that the generalized algorithm is furnished with the two distinctive properties for data points that are far from all centroids and for those that are close to any centroid. k-Harmonic means clustering are reviewed from the view point of FCM clustering. Graphical comparisons of the four classification functions are presented