摘要
对于维数较多的研究对象,为了研究的方便,总希望先将维数降低.主分量分析(PCA)和Fisher线性判别分析(FDA)是广泛应用于模式识别各个领域的两种常用方法.本文先利用主分量分析,将原始数据维数降低,然后再利用Fisher线性判别分析将维数再次降低,得到低维的数据,实验结果表明了两种方法结合的有效性.
As to an object of multi dimension, we always hope to make its dimension reduced in order to facilitate the study on it. Principal component analysis and Fisher linear diseriminant analysis are two common methods widely used in various fields of pattern recognition. This article reduced the dimension of original data, by the prineipal component analysis at first, and then use Fisher linear discriminant analysis to reduce the dimension once again, obtaining lower-dimensional data, finally experimental results demonstrated the effeetiveness of two methods' combination.
出处
《科技视界》
2015年第13期52-52,54,共2页
Science & Technology Vision
基金
重庆师范大学涉外商贸学院校级科研项目(KY2014006)
关键词
主分量分析
FISHER线性判别
距离判别法
Principle compoment analysis
Fisher linear discriminant analysis
A method of differentiating distances