摘要
现有的多模块多对多联想记忆模型结构复杂,不能够实现多粒度联想。在一对多联想模型基础上,加入项编码逻辑运算网络和粒度控制网络,提出了一种基于模式关联的多模块多对多联想记忆神经网络模型。实例表明,该模型可以实现模式的多对多联想记忆,且能够对输出模式进行粒度控制。从而实现了多粒度多对多联想记忆。
The structures of present and these models can not carry out associative memory model based on multimodule many to-many associative memory models are complex, the multi-granularity association. A kind of multimodule many-to-many incidence of patterns is proposed by use of introducing logic calculation network and granularity control network based on one-to-many associative memory model. Examples show that this model can achieve many-to-many associative memory and control the granularity of output patterns. Thus this model realizes multi-granularity and many-to-many associative memory.
出处
《青岛大学学报(工程技术版)》
CAS
2009年第4期12-18,24,共8页
Journal of Qingdao University(Engineering & Technology Edition)
基金
国家自然科学基金资助项目(60673101
60375014)
国家"八六三"高技术研究发展计划项目(2006AA01Z123
2006AA04Z110)
山东省自然科学基金资助项目(Y2007G30)
山东省科技攻关项目(2007GG3WZ04016)
关键词
多粒度联想
多对多联想记忆
模式编码
人工神经网络
many granularity association
many-to-many associative memory
pattern coding
artificialneural network