| 基于遗传算法的SVM多分类决策树优化算法研究 |
| |
| 引用本文： | 连可,陈世杰,周建明,龙兵,王厚军.基于遗传算法的SVM多分类决策树优化算法研究[J].控制与决策,2009,24(1). |
| |
| 作者姓名： | 连可 陈世杰 周建明 龙兵 王厚军 |
| |
| 作者单位： | 电子科技大学,自动化工程学院,成都,610054 |
| |
| 基金项目： | 国防科技应用基础研究基金 |
| |
| 摘 要： | 设计一种基于遗传算法(GA)的支持向量机(SVM)多分类决策树优化算法,以克服因传统SVM多分类决策树结构固定,单个SVM节点在树中位置随意而引起"误差积累"现象严重的缺陷.采用了SVM分类间隔作为GA适应度函数.利用GA在每一决策节点自动选择最优或近优的分类决策,最终自适应地实现了对决策树的优化.仿真实验表明,与传统方法相比,所提出的方法可使"误差积累"现象明显降低,分类质量大大提高.
|
| 关 键 词： | 支持向量机 遗传算法 决策树 误差积累 |
| 收稿时间： | 2007-10-19 |
| 修稿时间： | 2008-1-15 |
Study on GA-based SVM multi-class classification decision-tree optimization algorithm |
| |
| LIAN Ke,CHEN Shi-jie,ZHOU Jian-ming,LONG Bing,WANG Hou-jun.Study on GA-based SVM multi-class classification decision-tree optimization algorithm[J].Control and Decision,2009,24(1). |
| |
| Authors: | LIAN Ke CHEN Shi-jie ZHOU Jian-ming LONG Bing WANG Hou-jun |
| |
| Affiliation: | College of Automation Engineering;University of Electronic Science and Technology of China;Chengdu 610054;China. |
| |
| Abstract: | For the fixed tree configuration of traditional support vector machine(SVM) multi-class classification decision-tree algorithms and the random positions of their decision nodes,the error accumulate is very severity.Therefore,we present a GA-based SVM multi-class classification decision-tree optimization algorithm.We adopt the "margin" of SVM as adaption function to design GA.Then,GA is used to create optimal or suboptimal decision-tree automatically,which makes the margin between two classes maximal at ever... |
| |
| Keywords: | Support vector machine Genetic algorithm Decision tree Error accumulation |
| 本文献已被 CNKI 万方数据 等数据库收录！ |
| 点击此处可从《控制与决策》浏览原始摘要信息 |
|
