关键词: K-近邻, 密度, 模糊C均值, 协同系数

Abstract: The collaboration coefficient of Collaboration Fuzzy C-Means（CFC） algorithm is always determined by priori knowledge and remains constant during collaboration stages, with an inadequate using of the collaborative relationship. In order to circumvent this limitation, a novel CFC algorithm with K-nearest neighbor method determined collaboration coefficient is developed. Firstly, fuzzy partition matrix and cluster prototypes of every sub data sets are computed by Fuzzy C-Means（FCM） algorithm. Secondly, the number of nearest neighbors is setting and density of the cluster prototypes is gained by K-nearest neighbor method, forming density matrix. Thirdly, it dynamically adjusts collaborative coefficient by the correlation of density matrix. Lastly, it clusters objects with dynamical collaborative coefficient. Examples are provided to demonstrate the rationality of collaboration coefficient and the performance of collaboration FCM algorithm.

Key words: K-nearest neighbor, density, Fuzzy C-Means algorithm, collaborative coefficient