A CONSTRUCTIVE METHOD FOR MULTIVARIATE FUNCTION APPROXIMATION BY MULTILAYER PERCEPTRONS

被引：59
作者：
GEVA, S
SITTE, J
机构：
[1] Faculty of Information Technology, Queensland University of Technology, Brisbane, Queensland, 4001
来源：
IEEE TRANSACTIONS ON NEURAL NETWORKS | 1992年 / 3卷 / 04期
关键词：
D O I：
10.1109/72.143376
中图分类号：
TP18 [人工智能理论];
学科分类号：
081104 ; 0812 ; 0835 ; 1405 ;
摘要：
Mathematical theorems establish the existence of feedforward multilayered neural networks, based on neurons with sigmoidal transfer functions, that approximate arbitrarily well any continuous multivariate function. However, these theorems do not provide any hint on how to find the network parameters in practice. This paper shows how to construct a perceptron with two hidden layers for multivariate function approximation. Such a network can perform function approximation in the same manner as networks based on Gaussian potential functions, by linear combination of local functions.
引用
收藏
页码：621 / 624
页数：4
相关论文
共 10 条
[1]   APPROXIMATION-THEORY AND FEEDFORWARD NETWORKS [J].
BLUM, EK ;
LI, LK .
NEURAL NETWORKS, 1991, 4 (04) :511-515
[2]   ORTHOGONAL LEAST-SQUARES LEARNING ALGORITHM FOR RADIAL BASIS FUNCTION NETWORKS [J].
CHEN, S ;
COWAN, CFN ;
GRANT, PM .
IEEE TRANSACTIONS ON NEURAL NETWORKS, 1991, 2 (02) :302-309
[3]  
Cybenko G., 1989, Mathematics of Control, Signals, and Systems, V2, P303, DOI 10.1007/BF02551274
[4]   ON THE APPROXIMATE REALIZATION OF CONTINUOUS-MAPPINGS BY NEURAL NETWORKS [J].
FUNAHASHI, K .
NEURAL NETWORKS, 1989, 2 (03) :183-192
[5]  
GEVA S, 1991, NOV P IJCNN SING, P2305
[6]   CONDENSED NEAREST NEIGHBOR RULE [J].
HART, PE .
IEEE TRANSACTIONS ON INFORMATION THEORY, 1968, 14 (03) :515-+
[7]   MULTILAYER FEEDFORWARD NETWORKS ARE UNIVERSAL APPROXIMATORS [J].
HORNIK, K ;
STINCHCOMBE, M ;
WHITE, H .
NEURAL NETWORKS, 1989, 2 (05) :359-366
[8]  
Lapedes A. S., 1988, NEURAL INFORM PROCES, P442
[9]   A GAUSSIAN POTENTIAL FUNCTION NETWORK WITH HIERARCHICALLY SELF-ORGANIZING LEARNING [J].
LEE, S ;
KIL, RM .
NEURAL NETWORKS, 1991, 4 (02) :207-224
[10]   NETWORKS FOR APPROXIMATION AND LEARNING [J].
POGGIO, T ;
GIROSI, F .
PROCEEDINGS OF THE IEEE, 1990, 78 (09) :1481-1497
1