LEARNING OF NEURAL NETWORKS APPROXIMATING CONTINUOUS-FUNCTIONS THROUGH CIRCUIT SIMULATOR SPICE-PAC DRIVEN BY SIMULATED ANNEALING

被引：1

作者：

BERTHIAU, G

DURBIN, F

HAUSSY, J

SIARRY, P

机构：

[1] CEA,SERV ELECTRON,F-91680 BRUYERES CHATEL,FRANCE

[2] ECOLE CENT PARIS,ELECTRON LAB,F-92295 CHATENAY MALABRY,FRANCE

来源：

INTERNATIONAL JOURNAL OF ELECTRONICS | 1994年 / 76卷 / 03期

关键词：

D O I：

10.1080/00207219408925940

中图分类号：

TM [电工技术]; TN [电子技术、通信技术];

学科分类号：

0808 ; 0809 ;

摘要：

Simulated annealing (SA) adapted to continuous variables is used to determine the synaptic coefficients of an analogue multilayer neural network, approximating any continuous function of one or several variables. The 'open' electrical simulator SPICE-PAC driven by SA produces a globally optimal set of synaptic weights, in a reasonable time and without requiring heavy and inaccurate gradient computations. We illustrate and improve our weights tuning strategy through simple examples.

引用

页码：437 / 441

页数：5

共 5 条

[1]

BERTHIAU G, 1992, UNPUB ACM T MATH SOF

[2] CONSTRUCTION OF NEURAL NETWORKS TO APPROXIMATE ARBITRARY CONTINUOUS-FUNCTIONS OF ONE VARIABLE [J].