摘要
介绍并分析了模糊C-均值聚类算法、基于核方法的模糊C-均值聚类算法以及硬聚类算法。将硬聚类算法和模糊聚类算法结合起来,利用硬聚类算法初始化聚类中心,有效的减少模糊聚类算法的迭代次数。针对海量数据处理问题,将改进后的算法并行化,有效地提高了数据处理速度和效率,并在分布式互联PC环境下进行了性能测试。测试结果表明,基于核方法的并行模糊聚类算法具有很好的规模增长性和加速比。
Fuzzy C-means clustering algorithms (FCM), Fuzzy C-means clustering algorithms based on kernel method (FKCM) and C-means clustering algorithms (CM) are introduced and studied. FKCM and CM are put together, using CM to initialize centroids of FKCM to reduce iterations efficiently. In order to resolve the large amount of data, FKCM are made parallel, which improved the rapidity and efficiency of FKCM. And on the distributed linked PC/workstation, the parallel clustering algorithm is implemented. The result shows that the parallel algorithm has good sizeup and speedup.
出处
《计算机工程与设计》
CSCD
北大核心
2008年第8期1881-1883,共3页
Computer Engineering and Design
基金
广东省自然科学基金项目(05011896)
广东省教育厅自然科学基金项目(Z03080)
关键词
并行
模糊聚类
核方法
分布式
加速比
parallel
fuzzy clustering
kernel method
distributed
speedup