基于条件正定核的判决分类算法研究  

A Study of Kernel Discriminant Classify Algorithms Based on CPD Kernels

作　　者：王源[1] 陈亚军[2] 蔡彪[1] 

机构地区：[1]西华师范大学微机应用研究所 [2]西华师范大学物理与电子信息学院,四川南充637002

出　　处：《计算机技术与发展》2006年第8期51-54,共4页Computer Technology and Development

基　　金：四川省教育厅重点项目(2004A102)

摘　　要：在非线性分类算法中,重要的技术之一是对于核技巧的使用,目前仍是尚未解决的问题。针对Mercer核约束条件强的特点,引入条件正定核函数(c.p.d.核)改善其约束条件。实验证明,对于KPCA和KDDA等判决分类方法,通过c.p.d.核的使用,仍能保持其固有分类性能,从而推广了此类方法的适用范围。One of the important techniques is the trick of kernels among the nonlinear classify algorithms. It is a problcm which cannot be solved completely till now. For the more rigid restriction conditions of Mercer kernels,substitute them with conditionally positive kernels. The latter has less conditions so as to develop its usage. Some experiments show that c. p. d kernels can be used for several classical nonlinear discriminsent classify algorithms such as KPCA and KDDA. As a result, it extends the scope of these classify algorithms.

关 键 词：非线性分类算法 核主元分析 核直接判决分析 条件正定核 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]