基于聚类算法的并行化研究

聚类是数据挖掘中重要组成部分,为了提高聚类的处理效率,将并行处理技术运用于k-means和PAM算法中,对k-means与PAM算法进行了改进。实验结果表明:并行k-means算法相对串行k-means算法有更好的执行效率;且k-means算法有比PAM算法更好的并行性和可扩展性。最后,该文提出和介绍了将并行技术引入谱聚类算法。

Research on Parallelizing Based on Clustering Algorithm

Cluster analysis is an important component of data mining, aiming at improving the executive efficiency of clustering. In this paper, a method of parallel operating is applied to k-means algorithm and PAM algorithm, in order to improve these two algorithms. Experiments show that: parallel k-means algorithm has better performance than serial k-means algorithm; and k-means algorithm has better parallelism and extendibility than PAM algorithm. Finally, this paper puts forward the idea of introducing the method...