摘要： 针对熵模糊聚类算法只考虑特殊的加权指数问题，将广义熵引入到模糊聚类的目标函数，获得一种基于广义熵的模糊聚类模型和模糊聚类算法。将核函数引入到该模糊聚类模型中，提出基于广义熵的核模糊聚类算法。实验研究广义熵模糊聚类算法与核模糊聚类算法，证明当使用熵模糊聚类算法对数据聚类时，选取加权指数大于2的值可获得较好的聚类结果，同时参数对核算法的聚类结果有较大的影响。

关键词: 广义熵, 加权指数, 目标函数, 核函数, 模糊聚类

Abstract: Aiming at entropy fuzzy clustering algorithm only dealing with specific weight exponents, a fuzzy clustering model is obtained by combining the generalized entropy with objective function in fuzzy clustering. On the basis of the model, fuzzy clustering algorithm based on generalized entropy is presented. Moreover, kernel function is introduced into fuzzy clustering model and kernel fuzzy clustering algorithm based on generalized entropy is obtained. Experiments are conducted with both fuzzy clustering algorithm based on generalized entropy and its kernel fuzzy clustering algorithm. Results show that when weighting exponent’s value is greater than two, good clustering results are obtained using entropy fuzzy clustering algorithm to clustering data. At the same time, parameters in kernel clustering algorithm have the great impact on clustering results.

Key words: generalized entropy, weight exponent, objective function, kernel function, fuzzy clustering

中图分类号: 

• TP301.6