摘要
针对传统K-means算法中对初始化聚类中心敏感,容易陷入局部极小值等缺点,提出了一种基于粒子群算法和多类合并方法的新型K-means聚类算法.该算法首先利用改进粒子群算法选取初始聚类中心,然后利用K-means算法进行优化聚类,最后根据多类合并条件进行聚类合并,以获取最佳聚类结果.实验结果证明,该算法能有效解决传统K-means算法存在的缺陷,具有更快的收敛速度及更好的全局搜索能力,聚类划分效果更优.
To deal with the problem of the sensitivity of initialization and premature convergence, this paper proposes a novel K-means optimization clustering algorithm based on particle swarm optimization and multi-groups merging, namely M-PSO-Means. Firstly the algorithm selects the initial cluster center by improving particle swarms clustering algorithm under default number of clustering, then optimizes the clustering, and last carries out cluster merging based on multi-groups merging condition to obtain the best clustering results. The experimental results show that, the algorithm can effectively solve the defects of K-means algorithm, and has a faster convergence rate and better global search ability, as well as better cluster category effect.
出处
《计算机系统应用》
2014年第2期160-165,69,共7页
Computer Systems & Applications
基金
浙江省教育厅科研项目(Y201326770)
浙江省教育厅科研项目(Y201326872)
宁波大学科研基金项目(XYL12009)
浙江省2011年度大学生新苗人才计划项目
关键词
粒子群算法
多类合并
K
MEANS算法
适应度方差
particle swarm optimization（PSO）
multi-groups merging
K-means algorithm
fitness variance
