摘要
从ＢＡＭ的能量函数出发推导出强化系数的解析解形式。在强化系数的作用下，ＢＡＭ网络将被强化矢量对存储在以此矢量对为中心Ｈｅｍｍｉｎｇ距离为１的邻域里的能量最小点，从而保证了该矢量对的正确联想．给出了根据此解析算式编制的算法和计算实例．
This paper mainly discusses the choice of the reinforcing parameter in multiple training BAM algorithm. Starting from the energy function of the BAM network, an explicit solution of the parameter is finally reached. With this parameter, the BAM network can store the corresponding training vector pair at the minimum energy point within the area of 1Hemming distance from the vector pair. Thus ensures the pair can be correctly recalled later. An algorithm based on the explicit solution and an example are also given.
出处
《上海交通大学学报》
EI
CAS
CSCD
北大核心
1996年第11期56-61,共6页
Journal of Shanghai Jiaotong University
基金
国家教委跨世纪人才基金
上海市科学技术基金
关键词
双向联想记忆
存储器
强化系统
解析解
bidirectional associative memory (BAM)
neural network
Hemming distance
