Kosko型双向联想记忆与Hopfield神经网络的关系  

Relation between Kosko Bidirectional Associative Memories and Hopfield Networks

作　　者：张建国[1] 殷勤业[1] 

机构地区：[1]西安交通大学电子与信息工程学院,陕西西安710049

出　　处：《电气电子教学学报》2011年第2期98-101,共4页Journal of Electrical and Electronic Education

摘　　要：深入理解Kosko型双向联想记忆(BAM)网络是学习使用其它双向联想记忆的基础。如果我们按照传统的方法介绍BAM网络,学生很容易产生两个疑问:能量函数为什么是定义的?BAM网络与Hopfield网络的内在联系是什么?本文在说明Kosko型BAM网络就是一种特殊的Hopfield网络后,从Hopfield网络的能量函数出发推导出BAM网络的能量函数,从而揭示了它与Hopfield网络之间的内在联系。Kosko Bidirectional Associative Memory （BAM） network is one of the most fundamental BAM networks. When it is introduced with the traditional description, students usually have two questions, why the energy function is given as a definition and what is the relation between Kosko BAM and Hopfield networks. In this paper it is shown at that first Kosko BAM network is just one type of Hopfield neural networks, and then the energy function of BAM is naturally acquired based on the energy function of Hopfield network. Thereby the relation of them is exposed.

关 键 词：HOPFIELD神经网络 双向联想记忆 能量函数 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]