多分类器融合中一种新的加权算法

被引:10
作者
赵谊虹
程国华
史习智
机构
[1] 上海交通大学图像通信与信息处理研究所
[2] 上海交通大学振动、冲击、噪声国家重点实验室 上海
[3] 上海
关键词
多分类器融合; 加权投票表决; 贝叶斯推理;
D O I
10.16183/j.cnki.jsjtu.2002.06.006
中图分类号
TN911.3 [调制理论];
学科分类号
081002 ;
摘要
提出了一种直接采用分类器的输出向量来计算各分类器的加权算法 ,它能直接利用在分类器的输出端提供的“测量级”信息 ,通过加权函数将“测量级”信息转化为对分类器的加权 .为了提高系统的可靠性 ,在实验中还分析了表决阈值的选取 .
引用
收藏
页码:765 / 768
页数:4
相关论文
共 7 条
[1]  
Computer recognition of totally unconstrained handwritten ZIP codes. Ahmed P,Suen C Y. Pattern Recognition . 1987
[2]  
Pattern recognition with neural networks in C ++. Abhijit S P,Robert B M. . 1994
[3]  
Soft combination of neural classifiers: a comparative study. Verikas A,Lipnickas A,Malmqvist K,et al. Pattern Recognition . 1999
[4]  
A combination of statistic and syntactical pattern recognition applied to classification of unconstrained handwritten numerals. Duerr B,Haettich W,Tropf H,et al. Pattern Recognition . 1980
[5]  
Optimal combining of pattern classifiers. Lam L,Suen C Y. Pattern Recognition . 1995
[6]  
Methods of combining multiple classifiers and their applications to handwriting recognition. Xu L,Krzyzak A,Suen C Y. IEEE Transactions on Systems Man and Cybernetics . 1992
[7]  
Application of majority voting to pattern recognition: an analysis of its behavior and performance. Lam L,Suen C Y. IEEE Transactions on Systems Man and Cybernetics . 1997