电信科学 ›› 2015, Vol. 31 ›› Issue (8): 78-83.doi: 10.11959/j.issn.1000-0801.2015161

• 研究与开发 • 上一篇    下一篇

超平面支持向量机简化性能分析

程辉1,方景龙1,王大全2,王兴起1   

  1. 1 杭州电子科技大学复杂系统建模与仿真教育部重点实验室 杭州 310018
    2 杭州电子科技大学信息工程学院 杭州 310018
  • 出版日期:2015-08-27 发布日期:2015-08-27
  • 基金资助:
    国防基础科研计划基金资助项目;国防技术基础科研计划基金资助项目;武器装备预研基金资助项目

Performance Analysis of Simplification of Hyperplane Support Vector Machine

Hui Cheng1,Jinglong Fang1,Daquan Wang2,Xingqi Wang1   

  1. 1 Education Ministry Key Laboratory of Complex Systems Modeling and Simulation,Hangzhou Dianzi University,Hangzhou 310018,China
    2 Information Engineering School,Hangzhou Dianzi University,Hangzhou 310018,China
  • Online:2015-08-27 Published:2015-08-27
  • Supported by:
    Defense Basic Research Program;Defense Technical Basic Research Program;Weapon Equipment Advance Research Program

摘要:

与传统支持向量机相比,针对复杂分类问题的超平面支持向量机和针对高噪声数据回归问题的回归型超平面支持向量机,具有支持向量少、测试速度快、计算精度高的优点。然而对不同的样本集,超平面支持向量机的简化效果有所不同,仔细分析了这一现象的原因,得出在支持向量中非约束支持向量所占比率越低则超平面支持向量机简化效果越明显的结论。

关键词: 模式识别, 超平面支持向量机, 简化支持向量机, 性能分析

Abstract:

Comparing with traditional support vector machine,hyperplane support vector machine (HPSVM)and hyperplane support vector machine for regression(HPSVMR)not only reduce the number of support vectors and calculation time but also have comparable accuracy.However,they have different simplification effect on different problems.The reasons for this phenomenon were analyzed,the conclusion that the lower the percentage of non-bound support vectors is,the more obvious the effect of simplification was pointed out.

Key words: pattern recognition, hyperplane support vector machine, reduced support vector machine, performance analysis

No Suggested Reading articles found!