The Parallel Support Vector Machine Based on Hadoop Cloud Platform
暂无分享,去创建一个
Support vector machine( SVM) has shown a good performance in solving the small sample,nonlinear and high dimensional pattern recognition problems. However,with the increase of the data set size in practical,the algorithm becomes very slow to find the global optimum support vector process,even can not get the training model under effective time and actual environmental conditions allowed. Combining with the currently popular distributed solutions-Hadoop cloud platform,it is efficient to design and implement a parallel algorithm of SVM. Analyzing by experiments on UCI standard data set,the results show that the training time complexity has been reduced more obviously comparing with stand-alone SVM algorithm,without significantly reducing the prediction accuracy of the premise.