TRAINING BINARY NODE FEEDFORWARD NEURAL NETWORKS BY BACK PROPAGATION OF ERROR

被引：18

作者：

TOMS, DJ

机构：

[1] TACAN Corporation, California, 2330 Faraday Avenue Carlsbad

来源：

ELECTRONICS LETTERS | 1990年 / 26卷 / 21期

关键词：

Adaptive systems and Control; Neural networks;

D O I：

10.1049/el:19901121

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：

Despite the absence of derivatives, binary node neural networks having a hidden layer and multiple outputs can be trained using an algorithm which closely resembles conventional back propagation. The algorithm is based on the use of hidden unit activation functions which transform in the course of the training from analogue (sigmoid) to binary (step). © 1990, The Institution of Electrical Engineers. All rights reserved.

引用

页码：1745 / 1746

页数：2