GLOBAL CONVERGENCE AND SUPPRESSION OF SPURIOUS STATES OF THE HOPFIELD NEURAL NETWORKS

被引：58

作者：

ABE, S

机构：

[1] Hitachi Research Laboratory, Hitachi, Ltd, Hitachi

来源：

IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS I-FUNDAMENTAL THEORY AND APPLICATIONS | 1993年 / 40卷 / 04期

关键词：

D O I：

10.1109/81.224297

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：

Assuming that the output function of neurons is monotonic and differentiable at any interior point in the output range, we clarify the condition necessary for a vertex of a hyper-cube to become a local minimum of the Hopfield neural networks and the form of the convergence region to that minimum. Based on this, we derive a method to analyze and suppress spurious states in the networks. Finally we show that all the spurious states of the TSP for the Hopfield original energy function can be suppressed by our method, and we demonstrate validity of the method by computer simulations.

引用

页码：246 / 257

页数：12