摘要：

针对Apriori算法在Hadoop平台下集群节点计算能力有限的问题,将并行能力较大的GPU与Hadoop相结合,提出一种GPU-Hadoop的计算结构算法。通过Hadoop平台的MapReduce框架,节点将Apriori算法的计算密集型任务交由GPU进行处理,以缩减运算时间。实验结果表明,改进Apriori算法在面对大规模数据集时具有较高的执行速度与计算效率。

关键词: Apriori算法, Hadoop平台, 集群节点, 密集型任务, 大规模数据集

Abstract:

Aiming at the problem that the Apriori algorithm has limited computing power in the Hadoop platform,the GPU with Hadoop is combined with Hadoop,and a computing structure of GPU-Hadoop is proposed.Through the MapReduce framework of the Hadoop platform,nodes deliver the computationally intensive tasks of the Apriori algorithm to the GPU for processing to reduce computation time.Experimental results show that the improved Apriori algorithm has better execution speed and computational efficiency in the face of large-scale dataset.

Key words: Apriori algorithm, Hadoop platform, cluster node, intensive task, large-scale dataset

中图分类号: 

• TP311