基于向量组的Fisher线性鉴别分析方法  

Fisher linear discriminant analysis algorithm based on vector muster

作　　者：朱明旱[1] 邵湘怡[1] 

机构地区：[1]湖南文理学院电气与信息工程学院,湖南常德415000

出　　处：《计算机工程与应用》2011年第6期205-207,215,共4页Computer Engineering and Applications

基　　金：湖南省教育厅科研项目(No.10B074);湖南文理学院博士启动项目

摘　　要：提出了一种基于向量组的Fisher线性鉴别分析方法。该方法先将原始的高维向量分割为低维的子向量组,再对向量组运用Fisher线性鉴别分析。这种处理方法,不但能够解决任意高维下的小样本问题,而且通过选择恰当的子向量维数,可以从向量中抽取出最有效的特征值。此外,基于向量组的Fisher线性鉴别分析是Fisher线性鉴别分析和二维Fisher线性鉴别分析的进一步推广。Fisher linear discriminant analysis algorithm based on vector muster is presented in this paper.The original high-dimensional vectors are divided into a set of sub-vectors with low-dimensional.Fisher linear discriminant analysis is adopted based on vector muster.This algorithm can deal with all high-dimensional and small sample size problems.Otherwise, select- ing appropriate dimension of sub-vector can extract the optimization feature value of vector.Fisher linear discriminant analysis algorithm based on vector muster is the extension of Fisher linear discriminant analysis and two-dimensional fisher linear discriminant analysis.

关 键 词：FISHER线性鉴别分析 类间散布矩阵 类内散布矩阵 高维小样本问题 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]