摘要
粒子群算法是一类高效求解连续函数优化的随机搜索算法,在K均值聚类算法中得到广泛应用,但是在群体进化后期容易陷入局部极值,针对算法缺点,提出了一个新的聚类算法——基于免疫过程的粒子群K均值聚类算法,并将此算法与K均值聚类算法和粒子群K均值聚类算法进行比较。理论分析和数据实验证明,该算法有较好的全局收敛性,不仅能有效的克服传统的K均值聚类陷入局部极小值的缺点,而且全局收敛能力优于基于粒子群的K均值聚类算法。
Particle Swarm Optimization （PSO） is an efficient global stochastic search algorithm for continuous function optimization,and is combined broadly in K-Means clustering algorithm. But both K- Means and PSO are getting into local extreme in the anaphase of their evolution. An artificial immunity based PSO algorithm is proposed to overcome this defect and is compared with both K-Means algorithm and PSO based K-Means algorithm. Both theory analysis and experiments indicate that the new algorithm has greater global convergence ability. It overcomes the flaw of getting into local extreme by traditional K-Means effectively,and outperforms the PSO based K-Means algorithm in global convergence ability.
出处
《广西师范大学学报（自然科学版）》
CAS
北大核心
2008年第3期165-168,共4页
Journal of Guangxi Normal University:Natural Science Edition
基金
国家自然科学基金资助项目(10571073)
关键词
K均值
聚类
粒子群
免疫
K-means
cluster
particle swarm optimization
immunity
