一种新的形态学联想记忆网络模型研究  

RESEARCH ON A NEW MORPHOLOGICAL ASSOCIATIVE MEMORY NETWORK MODEL

作　　者：谷海红[1] 李素娟[1] 敖连辉[2] 

机构地区：[1]鹤壁职业技术学院电子信息工程学院,河南鹤壁458030 [2]河南师范大学计算机与信息技术学院,河南新乡453007

出　　处：《计算机应用与软件》2014年第6期165-169,231,共6页Computer Applications and Software

基　　金：河南省自然科学基金项目(0511012300)

摘　　要：针对即使在输入模式无噪声,形态学联想记忆在用于异联想时仍不能保证完全回忆的问题,从扩大记忆矩阵的存储空间的角度入手,提出一种新的形态学联想记忆模型——三维存储矩阵的形态学联想记忆来刻画MAM(Morphological Associative Memories)的记忆性能。该模型能够弥补传统形态学联想记忆的记忆矩阵的不足,解决MAM在异联想时不能保证对模式对集实现完全回忆的问题。详细阐述了构建三维存储矩阵的原理与步骤,并通过实例验证三维存储矩阵的形态学联想记忆的记忆性能远远优于传统的形态学联想记忆。When the morphological associative memories （MAM） are used for hetero association, it may not be ensured the complete recall even if the input patterns are noiseless. In light of this problem, starting from the perspective of expanding the storage space of memory matrix,we propose a new morphological associative memory model, namely the morphological associative memory model with three-dimensional storage matrix to depict the memory performance of MAM. This model is able to compensate for the memory matrix insufficiency of traditional morphological associative memory matrix and to solve the problem of MAM that it can＇ t guarantee the complete recall on the mode of the set when in betero association. This paper elaborates the principle and steps of building a three-dimensional memory matrix, and verifies by examples that the memory performance of three-dimensional memory matrix is far superior to the traditional morphological associative memory.

关 键 词：形态学联想记忆 记忆性能 异联想 三维存储矩阵 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]