Study on Parallel Clustering Based on Asynchronous Communication

Abstract

In order to improve the speed of clustering data mining, aiming at the characteristics of BIRCH algorithm, this paper has done research and analysis on rapid clustering data mining in the cluster system, and presented several improvement suggestions such as using data parallel thinking, uniform data distribution strategy , clustering communication mode and the optimization of clustering results. Experiment results show that the paralleling algorithm using asynchronous communication can achieve better performance and speedup than that synchronous one.