计算机工程与应用 ›› 2008, Vol. 44 ›› Issue (22): 146-148.DOI: 10.3778/j.issn.1002-8331.2008.22.043

• 数据库、信号与信息处理 • 上一篇    下一篇

一种基于信息熵的多分类器动态组合方法

陈 冰,张化祥   

  1. 山东师范大学 信息科学与工程学院,济南 250014
  • 收稿日期:2007-10-10 修回日期:2008-01-21 出版日期:2008-07-11 发布日期:2008-07-11
  • 通讯作者: 陈 冰

Method of dynamic ensemble of multiple classifiers based on information entropy

CHEN Bing,ZHANG Hua-xiang   

  1. College of Information Science and Engineering,Shandong Normal University,Jinan 250014,China
  • Received:2007-10-10 Revised:2008-01-21 Online:2008-07-11 Published:2008-07-11
  • Contact: CHEN Bing

摘要: 为提高数据分类的性能,提出了一种基于信息熵[1]的多分类器动态组合方法(EMDA)。此方法在多个UCI标准数据集上进行了测试,并与由集成学习算法—AdaBoost,训练出的各个基分类器的分类效果进行比较,证明了该算法的有效性。

关键词: 多分类器, 信息熵, 聚类, 分类器组合, Adaboost

Abstract: A method of dynamic ensemble of multiple classifiers based on information entropy(EMDA) is proposed in the paper,in order to improve the classification performance of dataset.The algorithm is tested on the UCI benchmark data sets,and comparative classification efficiency with several member classfiers trained based on ensemble learning algorithm—Adaboost.In the end,the utility of EMDA algorithm can be proved in the paper.

Key words: multiple classifiers, information entropy, clustering, classifier ensemble, Adaboost