一种基于杜鹃搜索算法的聚类分析方法
A Clustering Approach Based on Cuckoo Search Algorithm
-
摘要: 受初始类中心的影响K-Means算法聚类结果容易陷入局部最优.基于遗传算法(Genetic Algorithm,GA)和粒子群优化算法(particle swarm optimization algorithm,PSO)的改进K-Means一定程度上改善了基本K-Means的性能,然而GA和PSO本身也容易陷入局部最优解.针对上述问题,提出一种新的聚类方法—基于杜鹃搜索算法(Cuckoo search algorithm,CS)的K-Means聚类方法,并将此算法与现有的基于GA的K-Means和基于PSO的K-Means进行比较.实验结果表明:该方法能有效地改善基本K-Means算法易陷入局部极值的缺点,而且全局寻优能力优于基于GA的K-Means和基于PSO的K-Means,是一种性能鲁棒的聚类方法.Abstract: Due to the influence of initial class centers K-Means algorithm is easy to fall into local optimal clustering result. K-Means based on Genetic Algorithm and particle swarm optimization algorithm could improve the performance of the basic K-Means in a certain extent, however, Genetic Algorithm and particle swarm optimization algorithm themselves are easy to fall into local optimal solution too. Thus, in the paper, a clustering method based on K-means clustering algorithm and cuckoo search algorithm is presented, which is able to overcome the drawbacks of basic K-means clustering algorithm. Further, the proposed approach has been compared with K-means clustering algorithm optimized by genetic algorithm and particle swarm optimization algorithm. The experimental results show that the method can effectively improve the shortcoming of k-means algorithm, which is a robust clustering method and has better global search capability in comparison with K-means clustering algorithm based on genetic algorithm and particle swarm optimization algorithm.