一种基于MapReduce的并行PSO-BP神经网络算法

发布时间：2018-07-21 17:33

【摘要】：为了提高BP神经网络算法的分类准确率和运行时间效率,利用PSO算法和并行化设计的思想,提出了Hadoop平台下基于MapReduce的PSO优化BP神经网络的并行化设计及实现方法。利用PSO算法优化BP神经网络的初始权值和阈值,提高算法分类准确率;采用MapReduce并行编程模型实现算法的并行化处理,解决了BP神经网络在处理大规模样本数据集时存在的硬件开销和通信开销大的问题。选用SUN Database场景图像库构造了5个不同规模的数据集,通过与传统的串行PSO-BP神经网络算法实验对比,并行化的PSO-BP神经网络算法分类准确率达92%左右,系统效率在0.85左右,在处理大规模数据集时具有明显的优越性。
[Abstract]:In order to improve the classification accuracy and running time efficiency of BP neural network algorithm, the parallel design and implementation method of PSO optimization BP neural network based on MapReduce under Hadoop platform is proposed by using PSO algorithm and parallel design idea. The PSO algorithm is used to optimize the initial weights and thresholds of BP neural network to improve the classification accuracy of the algorithm, and the parallel programming model of MapReduce is used to realize the parallelization of the algorithm. The problem of large hardware overhead and communication overhead of BP neural network in dealing with large sample data sets is solved. Five data sets of different scales are constructed by using sun Database scene image database. By comparing with the traditional serial PSO-BP neural network algorithm, the classification accuracy of parallel PSO-BP neural network algorithm is about 92%, and the system efficiency is about 0.85. It has obvious advantages in dealing with large data sets.
【作者单位】：忻州师范学院计算机科学与技术系;太原科技大学计算机科学与技术学院;
【基金】：2014年度山西省大学生创新创业训练项目(2014383) 2013年度国家自然基金项目(61202163) 2013年度山西省自然基金项目(2013011017-2)
【分类号】：TP183

【相似文献】