基于退火的状态转移式模糊C-均值聚类算法

传统的模糊C -均值 (FCM)聚类是一种基于梯度下降的优化算法 ,该方法对初始化较敏感 ,且易陷入局部极小。借用模拟退火中系统温度的概念 ,提出基于退火的状态转移式模糊C -均值聚类方法 ,利用温度来控制类别中心的选取概率。实验结果表明 ,该方法可克服系统对数据集及初始解的敏感度 ,且能尽量避免陷入局部极小 ,而获得质量更优的解。

An Annealing-Based State Transitional Fuzzy C-Mean Algorithm

The traditional fuzzy C-means(FCM) algorithm is an optimization algorithm based on gradient descending. it is sensitive to the initial condition and liable to be trapped in a local minimum. By borrowing the system temperature concept from the technique used in the simulated annealing, a state transitional fuzzy C-means(STFCM)algorithm is proposed, in which the probability of the selected clustering centers depends on the decreasing temperature. Experimental results show that the new algorithm can give us more satisfactory results by escaping from local minima regardless of the randomness of the data set and the selection of the initial clustering centers.;