摘要
针对模糊C均值(FCM)聚类算法具有初始聚类中心敏感和容易陷入局部最优的问题,提出了一种基于改进遗传算法(GA)的加权模糊C均值聚类算法,采用高斯变异算子,提高了遗传算法在每个峰值附近的局部搜索能力,用基于复相关系数的加权欧式距离代替欧式距离,改进了FCM算法的聚类目标函数。用改进的算法对国际标准测试数据Iris进行测试,实验结果表明改进后的算法具有更好的稳定性和健壮性,提高了聚类的效果。
In accordance with the problems that Fuzzy C-Mean （FCM） algorithm has sensitivity against the initial value and it is easy to trap in local optimum for the clustering results, an improved Genetic Algorithm （GA） based on weighted FCM clustering algorithm was proposed by using Ganssian mutation operator to improve the GA local search capabilities around every peak value. Multiple correlation coefficient based on the weighted Euclidean distance instead of Euclidean distance was used to improve the objective function of the FCM clustering algorithm. The improved algorithm was used to test the international standard Iris data. The experimental results show that the improved algorithm has better stability and robustness, and enhances the clustering effect.
出处
《计算机应用》
CSCD
北大核心
2009年第B12期260-262,共3页
journal of Computer Applications
关键词
模糊C均值
遗传算法
复相关系数
加权欧式距离
高斯变异算子
Fuzzy C-Mean （FCM）
Genetic Algorithm （GA）
multiple correlation coefficient
weighted Euclidean distance
Gaussian mutation operator
