GNRFCM: A new fuzzy clustering algorithm and its application

Abstract

A new fuzzy clustering algorithm is presented in the paper. The main defect of the traditional fuzzy partitional clustering algorithm is to know the number of clustering in advance and the limitation of FCM algorithm is only applicable to spheroid and sensitive to the isolated data. GNRFCM algorithm adds a weight to the membership of the data, which is to decrease the effect of the isolated data on the initial cluster center. Since the number of clustering and the initial cluster centers affect cluster result, this paper adopts a density function algorithm to find the initial cluster centers. A new cohesion formula is put forward in GNRFCM algorithm. The lines (or columns) of the membership matrix of GNRFCM algorithm are sorted and transformed; the membership matrix is blocked to realize hierarchical clustering. The experiment shows the precision and the efficiency of clustering GNRFCM algorithm is higher than traditional partitional clustering FCM algorithm.