J4 ›› 2010, Vol. 37 ›› Issue (1): 136-141.doi: 10.3969/j.issn.1001-2400.2010.01.024

• 研究论文 • 上一篇    下一篇

多源性数据SVM集成算法研究

常甜甜1;刘红卫1;冯筠2   

  1. (1. 西安电子科技大学 理学院,陕西 西安  710071;
    2. 西北大学 信息技术学院,陕西 西安  710069)
  • 收稿日期:2009-04-13 出版日期:2010-02-20 发布日期:2010-03-29
  • 通讯作者: 常甜甜
  • 作者简介:常甜甜(1981-),女,西安电子科技大学博士研究生,E-mail: changtiantian@gmail.com.
  • 基金资助:

    国家自然科学基金资助项目(60603098);陕西省教育厅科学研究计划项目资助项目(07JK381)

Support vector machine ensemble learning algorithm research based on heterogeneous data

CHANG Tian-tian1;LIU Hong-wei1;FENG Jun2   

  1. (1. School of Science, Xidian Univ., Xi'an  710071, China;
    2. School of Information Sci. and Tech., Northwest Univ., Xi'an  710069, China)
  • Received:2009-04-13 Online:2010-02-20 Published:2010-03-29
  • Contact: CHANG Tian-tian

摘要:

针对数据特征的多源性特点,提出基于分组特征支持向量机集成算法.该方法将特征分组,对不同组特征采用不同的核函数映射到高维空间后用支持向量机分类,最后采用投票的方法得出决策标记,所得到的成员分类器具有较高的差异性.与传统的集成方法相比,该方法具有较好的检测性能.

关键词: 集成学习, 支持向量机, 多源性, 医学图像

Abstract:

An SVM ensemble learning algorithm based on grouped features is proposed for heterogeneous data. The feature is grouped and trained with different SVM classifiers, and then the final predict labels are obtained by the voting method. The diversity component classifiers with higher classification performance are obtained. Experimental results show that, compared with traditional ensemble learning, this method has the best performance.

Key words: ensemble learning, support vector machine, heterogeneous, medical image