Multi-central general type-2 fuzzy clustering approach for pattern recognitions

This paper presents a new approach to general type-2 fuzzy clustering called Multi-central general type-2 fuzzy clustering model. This model mainly focuses on uncertainty associated with the cluster centers. In this model, a set of points is considered as the center for each cluster. The membership values to the clusters are defined as general type-2 fuzzy sets including primary and secondary variables. Primary variable indicates the degree of belonging to the central objects, and the secondary variables indicate the degree of belonging of the central objects to the center of the cluster. There is not any type reduction or defuzzification for updating cluster prototypes in the proposed clustering algorithm. The compatible indexes with the proposed model are defined for validation and verification of the clustering process and results. Several experimental results are given to evaluate the performance of the proposed Multi-central general type-2 fuzzy clustering. The results also compared with the results of the type-1 fuzzy clustering model.