摘要
针对最小二乘支持向量机处理大规模数据集耗时长且受内存限制的特点,将局部多模型方法与MapReduce编程模式相结合,提出一种并行最小二乘支持向量机回归模型。模型由两组MapReduce过程组成,首先按照输入样本集对样本数据进行聚类操作,再对聚类后得到的子类按输出样本集进行二次聚类操作,分别得到局部模型数目和各局部模型综合加权输出计算结果。实验结果表明,并行最小二乘支持向量机回归模型具有较好的加速比和可扩展性。
According to the characteristics of least squares support vector machine regression model for long processing time and memory constraints,this paper designed a parallel least squares support vector machine regression model based on MapReduce and local multi-model method. The model was composed of two MapReduce process. It clustered the sample data according to the input set,and then obtained second clustering after sub set according to the output. Two MapReduce processes were calculated the number of local model and weighted output of each model. Experimental results show that the proposed parallel least squares support vector machine regression model has better speedup and scaleup.
出处
《计算机应用研究》
CSCD
北大核心
2015年第4期1060-1064,共5页
Application Research of Computers
基金
国家自然科学基金资助项目(51308057,51378073)
中国博士后科学基金面上资助项目(2014M550475)
国家教育部创新团队发展计划资助项目(IRT1050)
交通运输部基础研究基金资助项目(2010-319-812-080)
陕西省自然科学基础研究计划资助项目(2014JQ8354)
中央高校基本科研业务费专项资金资助项目(0009-2014G1321041,2013G3324005)
