Generalization Property of Fuzzy Classification Function for Tsallis Entropy-Regularization of Bezdek-Type Fuzzy C-Means Clustering.

In this study, Tsallis entropy-regularized Bezdek-type fuzzy c-means clustering method is proposed. Because the proposed method reduces to four conventional fuzzy clustering methods by appropriately controlling fuzzification parameters, the proposed method is considered to be their generalization. Through numerical experiments, this generalization property is confirmed; in addition, it is observed that the fuzzy classification function of the proposed method approaches a value equal to the reciprocal of the cluster number.