复形态双向联想记忆网络及其性能分析  

Complex Morphological Bidirectional Associative Memory Network and Its Performance Analysis

作　　者：徐迎[1] 曾水玲[2] 伍文源 

机构地区：[1]吉首大学物理与机电工程学院,湖南吉首416000 [2]吉首大学信息科学与工程学院,湖南吉首416000

出　　处：《信息与控制》2015年第3期270-275,共6页Information and Control

基　　金：国家自然科学基金资助项目(61363033;61363073);吉首大学科研资助项目(13JDY013;14JDY041)

摘　　要：基于复数环(C,∨,∧,+)提出了复形态双向联想记忆网络模型,给出了神经网络实现双向完全联想记忆的条件,并分析了自联想的存储能力、稳定性和收敛性及神经网络的抗噪声能力.当复数虚部取零时即为实数,实数域形态双向联想记忆网络就是复数域形态双向联想记忆网络的特例.最后通过仿真实验验证复形态双向联想记忆网络的有效性.A model of complex morphological bidirectional associative memory （ CMBAM ） based on the complex rings （ C,∨,∧,＋ ） is proposed. We gives the condition network realizing bidirectional perfect recall_ by using the neural network, analyzes storage capabilities, stability and convergence of self association, as well as the noise immunity of the neural network. Provided that the imaginary part of a complex number is equal to zero, the number is real Obviously, the real_ morphological, bidirectional, associative memories are a special case of the CMBAM. Lastly, the validity of CMBAM is proved through a simulation experiment.

关 键 词：复形态双向联想记忆网络 稳定性 收敛性 

分 类 号：TP273.2[自动化与计算机技术—检测技术与自动化装置]