【摘要】 研究模糊加权指数m对FCM(Fuzzy c-means)算法的聚类性能的影响,从划分熵入手提出了变权划分熵的概念,并基于模糊决策理论提出了一种最优加权指数m*的选取方法。该方法利用小的目标函数值和小的变权划分熵对应好的数据分类结果这一特性,将m的确定转化为一个带约束的非线性规划问题,从而确定最佳取值m*。实验结果表明该方法是非常有效和灵敏的。更多还原

【Abstract】 In this paper we study the effect of fuzzy weighted exponent m on the clustering performance of FCM(Fuzzy c-means)algorithm.It puts forward a concept of variable weight partition entropy based on partition entropy,and presents a method for choosing optimal weighted exponent m based on fuzzy decision theory.This method translates the determination of m into a problem of non-linear planning with restriction by utilising the characteristic of that the minimum objective function value and the minimum variable weight partition entropy correspond to the optimal results of data classification,and then the optimal value of m can be determined.The experimental results demonstrate that the proposed approach is very effective and sensitive.更多还原