Feature analysis: Neural network and fuzzy set theoretic approaches

被引:66
作者
De, RK [1 ]
Pal, NR [1 ]
Pal, SK [1 ]
机构
[1] INDIAN STAT INST,MACHINE INTELLIGENCE UNIT,CALCUTTA 700035,W BENGAL,INDIA
关键词
feature ranking; feature selection; neural networks; fuzziness measure;
D O I
10.1016/S0031-3203(96)00190-2
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
In this paper a new scheme of feature ranking and hence feature selection using a Multilayer Perceptron (MLP) Network has been proposed. The novelty of the proposed MLP-based scheme and its difference from another MLP-based feature ranking scheme have been analyzed. In addition we have modified an existing feature ranking/selection scheme based on fuzzy entropy. Empirical investigations show that the proposed MLP-based scheme is superior to the other schemes implemented. (C) 1997 Pattern Recognition Society. Published by Elsevier Science Ltd.
引用
收藏
页码:1579 / 1590
页数:12
相关论文
共 17 条
[1]  
[Anonymous], 1982, Pattern recognition: A statistical approach
[2]   DETERMINING INPUT FEATURES FOR MULTILAYER PERCEPTRONS [J].
BELUE, LM ;
BAUER, KW .
NEUROCOMPUTING, 1995, 7 (02) :111-121
[3]  
BHATTACHARJEE A, 1986, THESIS U CALCUTTA CA
[4]  
CHAKRABORTY B, 1994, THESIS CALCUTTA U CA
[5]   The use of multiple measurements in taxonomic problems [J].
Fisher, RA .
ANNALS OF EUGENICS, 1936, 7 :179-188
[6]  
JOHNSON RA, 1982, APPLIED MULTIVARIATE
[7]   INCORPORATING FUZZY MEMBERSHIP FUNCTIONS INTO THE PERCEPTRON ALGORITHM [J].
KELLER, JM ;
HUNT, DJ .
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 1985, 7 (06) :693-699
[8]   DEVELOPING HIGHER-ORDER NETWORKS WITH EMPIRICALLY SELECTED UNITS [J].
KOWALCZYK, A ;
FERRA, HL .
IEEE TRANSACTIONS ON NEURAL NETWORKS, 1994, 5 (05) :698-711
[9]   MEASURING FUZZY UNCERTAINTY [J].
PAL, NR ;
BEZDEK, JC .
IEEE TRANSACTIONS ON FUZZY SYSTEMS, 1994, 2 (02) :107-118
[10]   FUZZY SET THEORETIC MEASURE FOR AUTOMATIC FEATURE EVALUATION [J].
PAL, SK ;
CHAKRABORTY, B .
IEEE TRANSACTIONS ON SYSTEMS MAN AND CYBERNETICS, 1986, 16 (05) :754-760