您的位置:首页  > 论文页面

基于分解技术的并行支持向量机算法

发表时间:2013-07-15  浏览量:1736  下载量:673
全部作者: 李明强,韩丛英,贺国平
作者单位: 山东科技大学信息科学与工程学院;中国科学院大学数学科学院;山东科学院
摘 要: 针对大规模支持向量机(support vector machine, SVM)问题,提出基于分解技术的并行算法。首先介绍此算法的基本框架和涉及的并行计算技巧,然后给出一个新的子问题,并提出用基于一阶信息的加速方法来求解新的子问题,最后给出基于无梯度计算的工作集选取方式,并说明了这种方式的可行性。
关 键 词: 非线性规划;并行分解算法;一阶加速方法;支持向量机
Title: Parallel support vector machine algorithms based on decomposition technique
Author: LI Mingqiang, HAN Congying, HE Guoping
Organization: School of Information Science and Engineering, Shandong University of Science and Technology; School of Mathematical Sciences, University of Chinese Academy of Sciences; Shandong Academy of Sciences
Abstract: The parallel algorithm based on decomposition technique is proposed to solve the large scale support vector machine problems. Firstly, the outline of this algorithm and the skills in parallel computing are introduced. Then a new subproblem is given and the accelerated method based on first order information is proposed to solve the new subproblem. Finally, the working set selection method without gradient computation and its feasibility is described.
Key words: nonlinear programming; parallel decomposition algorithms; accelerated method based on first order information; support vector machine
发表期数: 2013年7月第13期
引用格式: 李明强,韩丛英,贺国平. 基于分解技术的并行支持向量机算法[J]. 中国科技论文在线精品论文,2013,6(13):1249-1254.
 
0 评论数 0
暂无评论
友情链接