Multiobjective GAs, quantitative indices, and pattern classification

被引:152
作者
Bandyopadhyay, S [1 ]
Pal, SK [1 ]
Aruna, B [1 ]
机构
[1] Indian Stat Inst, Machine Intelligence Unit, Kolkata 700108, India
来源
IEEE TRANSACTIONS ON SYSTEMS MAN AND CYBERNETICS PART B-CYBERNETICS | 2004年 / 34卷 / 05期
关键词
constrained elitism; genetic algorithms; hyperplane; fitting; minimal spacing; multiobjective optimization; (MOO); pareto-optimality; pattern recognition; purity;
D O I
10.1109/TSMCB.2004.834438
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 concept of multiobjective optimization (MOO) has been integrated with variable length chromosomes for the development of a nonparametric genetic classifier which can overcome the problems, like overfitting/overlearning and ignoring smaller classes, as faced by single objective classifiers. The classifier can efficiently approximate any kind of linear and/or nonlinear class boundaries of a data set using an appropriate number of hyperplanes. While designing the classifier the aim is to simultaneously minimize the number of misclassified training points and the number of hyperplanes, and to maximize the product of class wise recognition scores. The concepts of validation set (in addition to training and test sets) and validation functional are introduced in the multiobjective classifier for selecting a solution from a set of nondominated solutions provided by the MOO algorithm. This genetic classifier incorporates elitism and some domain specific constraints in the search process, and is called the CEMOGA-Classifier (constrained elitist multiobjective genetic algorithm based classifier). Two new quantitative indices, namely, the purity and minimal spacing, are developed for evaluating the performance of different MOO techniques. These are used, along with classification accuracy, required number of hyperplanes and the computation time, to compare the CEMOGA-Classifier with other related ones.
引用
收藏
页码:2088 / 2099
页数:12
相关论文
共 31 条
[1]  
[Anonymous], 1989, GENETIC ALGORITHM SE
[2]  
[Anonymous], 1987, PREPRINTS 31 ANN M I
[3]   VGA-classifier: Design and applications [J].
Bandyopadhyay, Sanghamitra, 2000, IEEE, Piscataway, NJ, United States (30)
[4]   The balance between proximity and diversity in multiobjective evolutionary algorithms [J].
Bosman, PAN ;
Thierens, D .
IEEE TRANSACTIONS ON EVOLUTIONARY COMPUTATION, 2003, 7 (02) :174-188
[5]  
Coello C, 1999, KNOWL INF SYST, V1, P129, DOI DOI 10.1007/BF03325101
[6]   Multiobjective optimization of trusses using genetic algorithms [J].
Coello, CA ;
Christiansen, AD .
COMPUTERS & STRUCTURES, 2000, 75 (06) :647-660
[7]   Constraint-handling using an evolutionary multiobjective optimization technique [J].
Coello, CAC .
CIVIL ENGINEERING AND ENVIRONMENTAL SYSTEMS, 2000, 17 (04) :319-346
[8]  
CONGALTON RG, 1983, PHOTOGRAMM ENG REM S, V49, P1671
[9]  
Deb K., 2000, Parallel Problem Solving from Nature PPSN VI. 6th International Conference. Proceedings (Lecture Notes in Computer Science Vol.1917), P849
[10]  
Deb K, 2002, IEEE C EVOL COMPUTAT, P825, DOI 10.1109/CEC.2002.1007032