Decision boundary preserving prototype selection for nearest neighbor classification

被引:45
作者
Barandela, R
Ferri, FJ
Sánchez, JS
机构
[1] Inst Technol Toluca, Metepec 52140, Mexico
[2] Univ Valencia, Dept Informat, E-46100 Valencia, Spain
[3] Univ Jaume 1, Dept Llenguatges & Sistemes Informat, Castellon de La Plana 12071, Spain
关键词
Nearest Neighbor rule; size reduction; classification accuracy; consistent subset; decision boundaries;
D O I
10.1142/S0218001405004332
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
The excessive computational resources required by the Nearest Neighbor rule are a major concern for a number of specialists and practitioners in the Pattern Recognition community. Many proposals for decreasing this computational burden, through reduction of the training sample size, have been published. This paper introduces an algorithm to reduce the training sample size while preserving the original decision boundaries as much as possible. Consequently, the algorithm tends to obtain classification accuracy close to that of the whole training sample. Several experimental results demonstrate the effectiveness of this method when compared to other reduction algorithms based on similar ideas.
引用
收藏
页码:787 / 806
页数:20
相关论文
共 31 条
[21]   Color segmentation based on a light reflection model to locate citrus fruits for robotic harvesting [J].
Pla, F. ;
Juste, F. ;
Ferri, F. ;
Vicens, M. .
Computers and Electronics in Agriculture, 1993, 9 (01) :53-70
[22]   A unifying view on instance selection [J].
Reinartz, T .
DATA MINING AND KNOWLEDGE DISCOVERY, 2002, 6 (02) :191-210
[23]   ALGORITHM FOR A SELECTIVE NEAREST NEIGHBOR DECISION RULE [J].
RITTER, GL ;
WOODRUFF, HB ;
LOWRY, SR ;
ISENHOUR, TL .
IEEE TRANSACTIONS ON INFORMATION THEORY, 1975, 21 (06) :665-669
[24]   Prototype selection for the nearest neighbour rule through proximity graphs [J].
Sanchez, JS ;
Pla, F ;
Ferri, FJ .
PATTERN RECOGNITION LETTERS, 1997, 18 (06) :507-513
[25]   2 MODIFICATIONS OF CNN [J].
TOMEK, I .
IEEE TRANSACTIONS ON SYSTEMS MAN AND CYBERNETICS, 1976, 6 (11) :769-772
[26]  
TOUSSAINT G, 2002, P 34 S COMP STAT MON
[27]   A COUNTEREXAMPLE TO TOMEK CONSISTENCY THEOREM FOR A CONDENSED NEAREST-NEIGHBOR DECISION RULE [J].
TOUSSAINT, GT .
PATTERN RECOGNITION LETTERS, 1994, 15 (08) :797-801
[28]   NEAREST NEIGHBOR PROBLEMS [J].
Wilfong, Gordon .
INTERNATIONAL JOURNAL OF COMPUTATIONAL GEOMETRY & APPLICATIONS, 1992, 2 (04) :383-416
[29]   Reduction techniques for instance-based learning algorithms [J].
Wilson, DR ;
Martinez, TR .
MACHINE LEARNING, 2000, 38 (03) :257-286
[30]   Optimal reference subset selection for nearest neighbor classification by tabu search [J].
Zhang, HB ;
Sun, GY .
PATTERN RECOGNITION, 2002, 35 (07) :1481-1490