摘要
在模式分类问题中,利用Fisher准则及K-L变换将样本数据从高维特征空间映射到低维特征空间以提取特征;而SVM(支持向量机)引进核函数隐含的映射把低维特征空间中的样本数据映射到高维特征空间来实现分类。文章利用三种方法对鸢尾属植物数据集的分类进行仿真试验,并对仿真结果进行分析比较,给出了三种方法在模式分类应用中的异同以及他们之间的内在联系和区别。
In the problem of pattern classification,the Fisher discriminant criterion and Karhunen-Loeve(K-L) transform have been used to extract features and discriminant the data from high-dimensional space to low-dimensional space.The SVM (Support Vector Machine) classifies the data by mapping the vector from low-dimensional space to highdimensional space using kernel function.In this paper,after experimenting on the classification of the data of iris with above 3 methods and analyzing the results,we get the association and differences among these 3 methods in the application of pattern classification.
出处
《计算机工程与应用》
CSCD
北大核心
2006年第19期147-150,157,共5页
Computer Engineering and Applications
基金
国家重点基础研究发展计划资助项目(编号:2004CB619303)
西安交通大学自然科学基金资助项目
关键词
FISHER准则
K-L变换
SVM(支持向量机)
分类
映射
投影
Fisher discriminant criterion,K-L transform,Support Vector Machine (SVM),classification,mapping,vector projection