|Table of Contents|

A Classifier Design Method Based on Sub class Division(PDF)

《南京理工大学学报》(自然科学版)[ISSN:1005-9830/CN:32-1397/N]

Issue:
1999年04期
Page:
293-296
Research Field:
Publishing date:

Info

Title:
A Classifier Design Method Based on Sub class Division
Author(s):
JingXiaoyuan JinXiaoyuan JinZhong YangJingyu
Department of Computer Science and Technology,NUST,Nanjing 210094
Keywords:
pat tern recognit ion face recog nition typical samples sub-class division
PACS:
TP391
DOI:
-
Abstract:
This paper gives a classif ier design method based on sub-class division, brings forward a new divisibility rule based on the w ithin-class scat ter matrix and the between-class scat ter mat rix and a method to ascertain the number of sub-class by t raining w ith the genetic a-l gorithm. Ex periments on a face database show that the new classif ier design method can improve the classificat ion ef fect of the neighbour method, and can solve the dif ficult problem of ascertaining just ly the typical samples in the classificat ion of the learning vector quant izat ion.

References:

1 黄德双. 神经网络模式识别系统理论. 北京: 电子工业出版社, 1996
2 边肇祺. 模式识别. 北京: 清华大学出版社, 1988
3 Leechter Yao. Nonlinear parameter estimatio n v ia the genetic algor ithm. IEEE Trans SP, 1994,42( 4) : 927~ 935
4 Rudolph G. Convergence analysis of cano nical genetic algor ithm. IEEE Trans NN, 1994, 5( 1) : 96~ 101
5 陈国良. 遗传算法及其应用. 北京: 人民邮电出版社, 1996
6 洪子泉, 杨静宇. 基于奇异值特征和统计模型的人像识别算法. 计算机研究与发展, 1994, 31( 3) : 60~ 65

Memo

Memo:
-
Last Update: 2013-03-29