| 一种快速SVM学习算法 |
| |
| 引用本文： | 张艳,兰光华,郁生阳,杨静宇. 一种快速SVM学习算法[J]. 计算机工程与应用, 2006, 42(32): 36-38 |
| |
| 作者姓名： | 张艳 兰光华 郁生阳 杨静宇 |
| |
| 作者单位： | 南京理工大学,计算机系,南京,210094;上海交通大学,图像处理与模式识别研究所,上海,200240 |
| |
| 摘 要： | 在对两种SVM学习算法(SMO和SVMlight)进行分析的基础上,提出了一种改进的基于集合划分和SMO的算法SDBSMO。该算法根据样本违背最优化条件的厉害程度将训练集划分为多个集合,每次迭代后利用集合信息快速更新工作集和相关参数,从而减少迭代开销,提高训练速度。实验结果表明该算法能很好地提高支持向量机的训练速度。
|
| 关 键 词： | 机器学习 支持向量机 学习算法 |
| 文章编号： | 1002-8331（2006）32-0036-03 |
| 收稿时间： | 2006-02-01 |
| 修稿时间： | 2006-02-01 |
Fast SVM Learning Algorithm |
| |
| ZHANG Yan,LAN Guang-hua,YU Sheng-yang,YANG Jing-yu. Fast SVM Learning Algorithm[J]. Computer Engineering and Applications, 2006, 42(32): 36-38 |
| |
| Authors: | ZHANG Yan LAN Guang-hua YU Sheng-yang YANG Jing-yu |
| |
| Affiliation: | 1.Dept. of Computer Science,Nanjing University of Science and Technology,Nanjing 210094,China; 2.Institute of Image Processing and Pattern Recognition,Shanghai Jiaotong University,Shanghai 200240,China |
| |
| Abstract: | Based on the analysis and comparision of two SVM training algorithms-SMO and SVM light,a revised algo-rithm(SDBSMO) which base on set-division and SMO is proposed.According to the violation degree of sample,it divides the training set into servral sets.After each iteration,SDBSMO makes use of the sets to update the working set and correlative parameters.Accordingly,the cost of each iteration is reduced so that SDBSMO can speed up the training.Finally,the experimental results show that it can substantially reduce the training time of SVM. |
| |
| Keywords: | machine learning Support Vector Machine(SVM) learning algorithm |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
|
