在分析传统分布式数据挖掘平台不足的基础上,结合网格服务的思想,提出了基于网格服务的分布式数据挖掘平台,同时在该平台上,实现了分布式BP网络分类算法(GBPC-GS)。仿真实验表明,与单机环境相比,随着网格节点数增加,算法的平均耗时明显下降,同时CPU的负载也下降了约40%。
This paper presents distributed data mining platform on grid services,which combines grid services with distributed data mining to solve problems of traditional data mining.Meanwhile,this paper implements Distributed BP Classification Algorithm on Grid Services(DBPC-GS).Simulation experiments show that contrasted with concentrative algorithms,the average consumptive time of DBPC-GS decreases apparently,and CPU load is reduced about 40% with increments of grid codes.