Parallel K-PSO based on MapReduce
Junjun Wang, Dongfeng Yuan, Mingyan Jiang · 2012
K-means is widely used in scientific research and commercial applications because of its simplicity and linearity. However, in faced of ever-growing amount of data and higher demand of cluster analysis today, how to improve the performance of K-means has become challenging and significant. So an improved method called parallel K-PSO which combines Particle Swarm Optimization (PSO) with K-means based on MapReduce is proposed in this paper. Firstly, it takes advantage of PSO to improve the global search ability of K-means, and then it makes K-means parallel with MapReduce to enhance its capability of processing massive data. We evaluate the proposed method through experimental results.