基于人工蜂群优化的K均值聚类算法  

A K-Means Clustering Algorithm Based on Artificial Bee Colony Optimization

作　　者：廖伍代[1] 朱范炳 王海泉[1] 孙雪凯 Liao Wudai;Zhu Fanbing;Wang Haiquan;Sun Xuekai(School of Electric and Information Engineering,Zhongyuan University of Technology,Zhengzhou 450007,China)

机构地区：[1]中原工学院电子信息学院,郑州450007

出　　处：《计算机测量与控制》2018年第4期136-138,156,共4页Computer Measurement &Control

摘　　要：为了改善K均值聚类算法对初始聚类中心敏感和易于陷入局部最优的不足,提出人工蜂群算法和K均值聚类算法相结合的想法,即基于人工蜂群优化的K均值聚类算法;通过全局寻优能力强的人工蜂群算法初始化K均值的聚类中心并优化聚类中心的位置,从而帮助K均值跳出局部极值,优化聚类效果;将混合聚类算法用Iris、Red Wine和New Red Wine数据集做聚类测试,结果表明该算法既克服了原始K均值聚类算法容易受初始聚类中心影响和不稳定的缺点,又具有良好的性能和聚类效果。In order to improve the shortcomings of K-Means algorithm,which are sensitive to initial clustering centers and easily caught in local optimum,proposes an idea that combines K-Means clustering algorithm with artificial bee colony algorithm.That is a K-Means clustering algorithm based on artificial bee colony optimization.With the strong ability of global optimization,the artificial bee colony algorithm can initialize the K-Means clustering centers in an effective way,and move the clustering centers to better positions.As a result of helping K-Means escape from local optimum and optimize clustering effect.Testing the hybrid clustering algorithm with UCI Iris,Red wine and New Red Wine data sets,results show that the algorithm not only overcomes instability of original K-Means,but also provides a better clustering performance.

关 键 词：聚类分析 K均值算法 人工蜂群算法 聚类中心 优化 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]