Characterizations of nearest and farthest neighbor algorithms by clustering admissibility conditions

被引：4

作者：

Chen, ZM ^{[1
]}

Van Ness, J

机构：

[1] Florida Int Univ, Dept Stat, Miami, FL 33199 USA

[2] Univ Texas, Program Math Sci, Richardson, TX 75083 USA

来源：

PATTERN RECOGNITION | 1998年 / 31卷 / 10期

关键词：

clustering; clustering admissibility; monotone clustering admissibility; nearest-neighbor clustering; farthest-neighbor clustering; Lance and Williams algorithms;

D O I：

10.1016/S0031-3203(98)00002-8

中图分类号：

TP18 [人工智能理论];

学科分类号：

081104 ; 0812 ; 0835 ; 1405 ;

摘要：

Monotone admissibility for clustering algorithms was introduced in Fisher and Van Ness [Biometrika 58, 91-104 (1971)]. The present paper discusses monotone admissibility for a broad class of clustering algorithms called the Lance and Williams algorithms. Necessary and sufficient conditions for Lance and Williams algorithms to be monotone admissible are discussed here. It is shown that the only such algorithms which are monotone admissible are nearest neighbor and farthest neighbor. (C) 1998 Published by Elsevier Ltd on behalf of the Pattern Recognition Society. All rights reserved.

引用

页码：1573 / 1578

页数：6

共 5 条

[1] SPACE-CONTRACTING, SPACE-DILATING, AND POSITIVE ADMISSIBLE CLUSTERING ALGORITHMS [J].