摘要
从划分聚类要求的时间和空间上看,传统的串行算法已很难适应海量的数据,有必要研发高性能、可扩展的并行算法来解决这一问题.基于一些主要的并行划分聚类算法所存在的问题,提出了在机群系统上采取数据并行策略设计的并行划分聚类算法思想.
It becomes nearly impossible to use serial partitional clustering algorithm to process high volumes of data, for both time and space reasons. There is a need to develop parallel methodology for this problem. Based on analyzing some existed parallel algorithms, this paper presented a parallel partition algorithm for PCs cluster, which adopts the idea of data parallelism.
出处
《西华师范大学学报（自然科学版）》
2005年第3期271-273,共3页
Journal of China West Normal University(Natural Sciences)
基金
西华师范大学科研启动基金资助项目(04B076)
关键词
并行划分聚类
机群
数据并行
负载平衡
parallel partitional clustering algorithm
PCs cluster
data parallelism
load balancing
