计算机工程与应用 ›› 2008, Vol. 44 ›› Issue (32): 140-142.DOI: 10.3778/j.issn.1002-8331.2008.32.042

• 数据库、信号与信息处理 • 上一篇    下一篇

改进的AdaBoost算法与SVM的组合分类器

李亚军,刘晓霞,陈 平   

  1. 西北大学 信息科学与技术学院,西安 710127
  • 收稿日期:2007-12-11 修回日期:2008-03-19 出版日期:2008-11-11 发布日期:2008-11-11
  • 通讯作者: 李亚军

Combined classification algorithm based on improved AdaBoost and SVM

LI Ya-jun,LIU Xiao-xia,CHEN Ping   

  1. Institute of Information Science and Technology,Northwest University,Xi’an 710127,China
  • Received:2007-12-11 Revised:2008-03-19 Online:2008-11-11 Published:2008-11-11
  • Contact: LI Ya-jun

摘要: 提出了一种改进的AdaBoost算法与支持向量机组合的分类方法,用来处理多类别分类。采用规则抽样来解决支持向量机分类中正负样本的不平衡性,改进AdaBoost算法,使其在初始化时考虑样本分布稀疏的重要性,有利于稀有类样本的正确划分。实验结果表明,此方法与标准支持向量机分类器相比,泛化性能有一定程度的提高。

关键词: AdaBoost, 支持向量机, 组合分类器, 规则抽样

Abstract: A combined classification algorithm based on improved AdaBoost and Support Vector Machine,is proposed in order to deal with the problems of multiclass classification.Adopt a rule sampling to solve the unbalance of samples in the SVM.Improving the AdaBoost makes it consider the importance of sparse sample distribution at the beginning,this is advantageous to the right demarcation of rare sample.Experiment proves this algorithm can raise the generalization ability compared with the standard SVM.

Key words: AdaBoost, Support Vector Machine(SVM), combined classification, rule sampling