Coevolution of nearest neighbor classifiers

被引:13
作者
Gagne, Christian [1 ]
Parizeau, Marc [1 ]
机构
[1] Univ Laval, Dept Genie Elect & Genie Informat, Lab Vis Syst Numeriques, Ste Foy, PQ G1K 7P4, Canada
基金
加拿大自然科学与工程研究理事会;
关键词
pattern recognition; nearest neighbor classification; prototype selection; proximity measure; evolutionary computation; genetic algorithms; genetic programming; coevolution; multiobjective optimization;
D O I
10.1142/S0218001407005752
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
This paper presents experiments of Nearest Neighbor (NN) classifier design using different evolutionary computation methods. Through multiobjective and coevolution techniques, it combines genetic algorithms and genetic programming to both select NN prototypes and design a neighborhood proximity measure, in order to produce a more efficient and robust classifier. The proposed approach is compared with the standard NN classifier, with and without the use of classic prototype selection methods, and classic data normalization. Results on both synthetic and real data sets show that the proposed methodology performs as well or better than other methods on all tested data sets.
引用
收藏
页码:921 / 946
页数:26
相关论文
共 46 条
[1]  
[Anonymous], 1998, Genetic programming: an introduction
[2]  
[Anonymous], 1999, GENETIC PROGRAMMING
[3]  
Back T., 2000, EVOLUTIONARY COMPUTA, V1
[4]  
BEAULIEU J, 2002, P 2002 GEN EV COMP C, P155
[5]  
Blake C.L., 1998, UCI repository of machine learning databases
[6]  
Bleuler S, 2001, IEEE C EVOL COMPUTAT, P536, DOI 10.1109/CEC.2001.934438
[7]  
Bot MCJ, 2001, LECT NOTES COMPUT SC, V2038, P256
[8]   Advances in instance selection for instance-based learning algorithms [J].
Brighton, H ;
Mellish, C .
DATA MINING AND KNOWLEDGE DISCOVERY, 2002, 6 (02) :153-172
[9]   Using evolutionary algorithms as instance selection for data reduction in KDD: An experimental study [J].
Cano, JR ;
Herrera, F ;
Lozano, M .
IEEE TRANSACTIONS ON EVOLUTIONARY COMPUTATION, 2003, 7 (06) :561-575
[10]   Design of nearest neighbor classifiers: multi-objective approach [J].
Chen, JH ;
Chen, HM ;
Ho, SY .
INTERNATIONAL JOURNAL OF APPROXIMATE REASONING, 2005, 40 (1-2) :3-22