WEIGHTED NEAREST NEIGHBOR RULE FOR CLASS DEPENDENT SAMPLE SIZES

被引：23

作者：

BROWN, TA ^{[1
]}

KOPLOWITZ, J ^{[1
]}

机构：

[1] CLARKSON COLL TECHNOL,DEPT ELECT & COMP ENGN,POTSDAM,NY 13676

来源：

IEEE TRANSACTIONS ON INFORMATION THEORY | 1979年 / 25卷 / 05期

关键词：

D O I：

10.1109/TIT.1979.1056092

中图分类号：

TP [自动化技术、计算机技术];

学科分类号：

0812 ;

摘要：

The nearest neighbor rule is considered for data samples obtained by selecting Niindependent samples with the conditional distribution corresponding to class Ci. It is shown that a weighted distance rule can improve the performance when the ratio of Nito the total sample size differs substantially from the a priori probability of class Ci. © 1979 IEEE

引用

页码：617 / 619

页数：3