NEURAL NETWORK CONTROLLER USING AUTOTUNING METHOD FOR NONLINEAR FUNCTIONS

被引：60

作者：

YAMADA, T

YABUTA, T

机构：

[1] NTT Telecommunication Field Systems R&D Center, Tokai-Mura, Ibaraki-ken

来源：

IEEE TRANSACTIONS ON NEURAL NETWORKS | 1992年 / 3卷 / 04期

关键词：

Mathematical Programming; Nonlinear--Applications; -; Optimization--Applications;

D O I：

10.1109/72.143373

中图分类号：

TP18 [人工智能理论];

学科分类号：

081104 ; 0812 ; 0835 ; 1405 ;

摘要：

Many studies, among them Kawato's work, have been undertaken in order to apply both the flexibility and the learning ability of neural networks to dynamic system controllers. However, many parameters, such as the optimum shape of the sigmoid functions (nonlinear saturated functions), are determined by trial and error. These parameters inhibit the advanced application of neural networks. Therefore, this paper proposes an autotuning method for the optimum sigmoid function of neural networks. Simulated results using a learning-type direct controller confirm both the practicality and the characteristics of our proposed autotuning method.

引用

页码：595 / 601

页数：7

共 17 条

[1]

Arai M., 1991, Transactions of the Institute of Electronics, Information and Communication Engineers A, VJ74A, P551

[2]

GUEZ A, 1987 P IEEE INT C NE, V4, P567

[3]

JORDAN MI, 1989, JUN P IJCNN89

[4]

KAWATO M, 1987 P IEEE INT C NE, V4, P573

[5]

KONDOU J, 1974, MARUZEN

[6]

NIZNIK CA, 1987 P IEEE INT C NE, V4, P583

[7]

PSALTIS D, 1987 P IEEE INT C NE, V4, P551

[8]

Rumelhart D. E., 1988, PARALLEL DISTRIBUTED

[9]

TOKITA T, 1989, J ROBOTICS SOC JAPAN, V7, P47

[10]

UAMADA Y, 1991, P IECON 91 INT C IND, P1389

← 1 2 →