一种基于ACO的K-medoids聚类算法  

K-medoids clustering algorithm method based on ant colony algorithm

作　　者：孟颖[1] 罗可[1] 姚丽娟[1] 王琳[1] 

机构地区：[1]长沙理工大学计算机与通信工程学院,长沙410114

出　　处：《计算机工程与应用》2012年第16期136-139,152,共5页Computer Engineering and Applications

基　　金：国家自然科学基金(No.10926189;No.10871031);湖南省自然科学衡阳联合基金(No.10JJ8008);湖南省教育厅重点项目(No.10A015)

摘　　要：K-medoids算法作为聚类算法的一种,不易受极端数据的影响,适应性广泛,但是K-medoids聚类算法的精确度不稳定,平均准确率较低,用于实际的聚类分析时效果较差。ACO是一种仿生优化算法,其具有很强的健壮性,容易与其他方法相结合,求解效率高等特点。在K-medoids聚类算法的基础上,借鉴ACO算法的优点,提出了一种新的聚类算法,它提高了聚类的准确率,算法的稳定性也比较高。通过仿真实验,验证了算法的可行性和先进性。K-medoids algorithm as a kind of clustering algorithm, not easily affected by extreme data, the influence of broad adaptability, but K-medoids clustering algorithm accuracy is not stable, average accuracy, low in the real,clustering analysis effect is poorer. ACO is a bionic optimization algorithm, which has strong robustness, can be unified easily with other method, has high efficiency. K-medoids clustering algorithm based on ACO algorithm merit reference, this paper proposes a new clustering algorithm. It raises the clustering algorithm, and the stability of the accuracy is high. Finally, simulation experiments show the feasibility and advantage of this algorithm.

关 键 词：蚁群优化算法(ACO) 聚类分析 K-medoids算法 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]