摘要
提出一种在大规模光电混合神经网络系统中实现可编程拓扑重构的有效方法。文中介绍了这种方法的原理。
An efficient method of programmable reconfiguration of the topology in an optical/digital neural system with large scale is proposed. The principle and application technique of the method are described. And the experimental results of operation of a single layer feedback network as well as a multi layer feedforward network in a NP1024 optical/digital neuroprocessor by using the proposed method are reported.
出处
《光学学报》
EI
CAS
CSCD
北大核心
1998年第10期1306-1310,共5页
Acta Optica Sinica
基金
国家教委博士点基金
国家自然科学基金资助的课题
关键词
光学神经网络
网络拓扑结构
可编程拓扑重构
optical neural networks, topology of neural network structures, reconfiguration of the topology.
