CASIA OpenIR  > 毕业生  > 硕士学位论文
基于再生核映射的分类器的研究与应用
邹冬方
Subtype工学硕士
Thesis Advisor刘迎建
2001-05-01
Degree Grantor中国科学院研究生院
Place of Conferral中国科学院自动化研究所
Degree Discipline模式识别与智能系统
Keyword核映射 分类器
Abstract多层感知器网络、径向基函数网络和支持向量机是三种经典的分类器,它 们比其它统计模式识别方法有较大的优势。而优势的根源在于它们实现了从输入 到输出的非线性映射。事实上,上述三种分类器所基于的非线性映射都可以归为再生核映射,此 映射可以在RKHS中利用线性空间中的方法来确定。 在对再生核映射及RKHS的概念进行深刻的理论讨论后,本文对另外一种 基于再生核映射的神经网络分类器-OI网进行了详细的说明,并在原设计者提 供的学习算法基础上提出了一种新的学习算法。 子空间法是一种现代统计模式识别方法。在本文中,子空间法中最基本的 方法一CLAFIC法与再生核映射被结合起来,从而形成一种新的分类器-RKHS 中的子空间分类器。局部子空间分类器是子空间法中最新的技术,它与再生核映 射相结合的产物一RKHS中的局部子空间分类器也由我们提出。
Other AbstractMLPN (Multilayer Perceptron Network), RBFN(Radial Basis Function Network) and SVM(Support Vector Machine) are three classical classifiers, which are superior to other statistical approaches. The superiority roots in the nonlinear mapping performed by them from the input vectors to the output ones. As a matter of fact, the nonlinear maps on which those classifiers are based are kinds of reproducing kernel maps, which can be formulated by linear methods in RKHS(Reproducing Kernel Hilbert Space). After the concepts of Reproducing Kernel Map and RKHS are discussed on basis of profound theories, another neural classifier based on Reproducing Kernel Map is demonstrated in particular, whose name is OI Network. The original learning algorithms for it given by the divisor are studied, and a new learning algorithm is put forward in this paper. Subspace Method is a modem pattern recognition method. In this paper, the most fundamental one of all subspace methods, named CLAFIC, is combined with Reproducing Kernel Map, then a new classifier, Subspace Classifier in Reproducing Kernel Hilbert Space, comes into being. Local Subspace Classifier is the newest technique of subspace methods, which is also combined with Reproducing Kernel Map, with the consequent product being Local Subspace Classifier in Reproducing Kernel Hilbert Space.
shelfnumXWLW604
Other Identifier604
Language中文
Document Type学位论文
Identifierhttp://ir.ia.ac.cn/handle/173211/6797
Collection毕业生_硕士学位论文
Recommended Citation
GB/T 7714
邹冬方. 基于再生核映射的分类器的研究与应用[D]. 中国科学院自动化研究所. 中国科学院研究生院,2001.
Files in This Item:
There are no files associated with this item.
Related Services
Recommend this item
Bookmark
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[邹冬方]'s Articles
Baidu academic
Similar articles in Baidu academic
[邹冬方]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[邹冬方]'s Articles
Terms of Use
No data!
Social Bookmark/Share
All comments (0)
No comment.
 

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.