摘要
本文提出了一种具有时空混沌控制的联想记忆网络 .实验结果表明 :具有目标信息的一部分知识的初始输入能在时空混沌的参数控制中成功地完成联想记忆 ,根据提出的学习算法 ,该网络的记忆搜索性能和记忆容量比Hopfield模型有较大改善 .同时发现联想记忆成功率与强化因子、样本数、信息率、学习阈值以及初始混沌参数有关 .
An associative memory network with spatiotemporal chaos control is investigated. Experimental results show that initial input with the knowledge of only a part of the target information can successfully complete associative memory by spatiotemporal chaos parameter control. By means of our proposed algorithm, memory search performance and memory capacity possess more powerful improvement compared with the Hopfield model. Meanwhile we find that success rate of associative memory depends on reinforcement factor, sampling number, information quantity, training threshold value and initial chaos parameters.
出处
《电子学报》
EI
CAS
CSCD
北大核心
2001年第5期678-681,共4页
Acta Electronica Sinica
基金
国家 8 63计划项目! (No .863 51 1 9845 0 0 2 )
国家自然科学基金! (No .60 0 750 0 8)
关键词
时空混沌控制
联想记忆网络
人工智能
神经网络
Algorithms
Automatic target recognition
Chaos theory
Computer simulation
Mathematical models
Sampling
Threshold logic
