FUNCTIONALLY EQUIVALENT FEEDFORWARD NEURAL NETWORKS

被引：40

作者：

KURKOVA, V ^{[1
]}

KAINEN, PC ^{[1
]}

机构：

[1] IND MATH, WASHINGTON, DC 20007 USA

来源：

NEURAL COMPUTATION | 1994年 / 6卷 / 03期

关键词：

D O I：

10.1162/neco.1994.6.3.543

中图分类号：

TP18 [人工智能理论];

学科分类号：

081104 ; 0812 ; 0835 ; 1405 ;

摘要：

For a feedforward perceptron type architecture with a single hidden layer but with a quite general activation function, we characterize the relation between pairs of weight vectors determining networks with the same input-output function.

引用

页码：543 / 558

页数：16

共 12 条

[1] ALBERTINI F, 1994, IN PRESS NEURAL NETW
[2] BIRKHOFF G, 1965, SURVEY MODERN ALGEBR
[3] ON THE GEOMETRY OF FEEDFORWARD NEURAL-NETWORK ERROR SURFACES
CHEN, AM
LU, HM
HECHTNIELSEN, R
[J]. NEURAL COMPUTATION, 1993, 5 (06) : 910 - 927
[4] CHEN AM, 1991, IEE CONF PUBL, V349, P1
[5] Daubechies I., 1992, 10 LECT WAVELETS, DOI 10.1137/1.9781611970104
[6] HECHTNIELSEN R, 1990, ADVANCED NEURAL COMPUTERS, P129
[7] APPROXIMATION CAPABILITIES OF MULTILAYER FEEDFORWARD NETWORKS
HORNIK, K
[J]. NEURAL NETWORKS, 1991, 4 (02) : 251 - 257
[8] KREINOVITCH V, 1992, UTEP92 TECH REP
[9] PATI YC, 1991, NEURAL INFORMATION P, V3, P743
[10] WAVELETS AND DILATION EQUATIONS - A BRIEF INTRODUCTION
STRANG, G
[J]. SIAM REVIEW, 1989, 31 (04) : 614 - 627

← 1 2 →