CONVERGENCE TIME ON THE RS MODEL FOR NEURAL NETWORKS

被引：7
|
作者：
Penna, T. J. P. [1 ]
de Oliveira, P. M. C. [1 ]
Arenzon, J. J. [2 ]
de Almeida, R. M. C. [2 ]
Iglesias, J. R. [2 ]
机构：
[1] Univ Fed Fluminense, Inst Fis, BR-24020 Niteroi, RJ, Brazil
[2] Univ Fed Rio Grande Sul, Inst Fis, BR-91500 Porto Alegre, RS, Brazil
来源：
INTERNATIONAL JOURNAL OF MODERN PHYSICS C | 1991年 / 2卷 / 03期
关键词：
Neural Networks; Multispin Coding; Multineuron Models;
D O I：
10.1142/S0129183191000950
中图分类号：
TP39 [计算机的应用];
学科分类号：
081203 ; 0835 ;
摘要：
Convergence times and the corresponding dispersions have been studied numerically as parameters to measure the efficiency of neural network models. These quantities are also supposed to be related to the number of spurious states for each configuration of stored patterns. In this work we measure these quantities for a recent multineuron interaction model presenting an enhanced performance compared to other traditional schemes.
引用
收藏
页码：711 / 717
页数：7
相关论文
共 50 条
12345