Hopfield神经网络模型和改进模型的理论分析及单通道光学实现  

Theoretical Analysis and Single Channel Optical Implementation of the Hopfield Neural Network Model and Its Improved Model

作　　者：吴平凡[1] 余飞鸿[1] 李正民[1] 

机构地区：[1]浙江大学光仪系信息处理研究室,杭州310027

出　　处：《半导体光电》1992年第3期248-251,264,共5页Semiconductor Optoelectronics

摘　　要：本文通过互补扩展存贮模式的方法,对 Hopfield 神经网络模型进行了改进,提高了网络的性能和存贮能力。同时,从理论上分析比较了 Hopfield 模型和互补扩展模型全正型实现方法,给出了两套全正型单通道光学实现系统。A modified Hopfield neural network model is presented in this paper.The performance and storage capability are improved by means of comple- mentary extention of the storage and input modes.In addition,theoretical analysis and comparison are made on all-positive connection weight optical implementation of Hopfield and complementary extention models.Two sets of all-positive single channel optical implementation systems are given.

关 键 词：神经网络 信息处理 光调制器 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]