摘要
报道用自行研制的ＣａＳ（Ｅｕ，Ｓｍ）电子俘获材料表示互联权重矩阵以实现光学ＩＰＡ神经网络模型。由于ＣａＳ（Ｅｕ，Ｓｍ）的红外激励发光强度与俘获的电子密度及红外读出光强度之积成线性关系，故可用来表示互联权重矩阵以实现光学神经网络。这种互联权重矩阵具有很宽的数值范围，并可以用光学的办法进行快速擦除。
Optical IPA(Interpattern association) neural network model is implemented using CaS(Eu, Sm) electron trapping materials synthesized by the authors. The intensity of infrared stimulated luminescence from CaS(Eu, Sm) shows a linear relationship with the product of the trapped electrons' density and the infrared reading power. Therefore, these materials are capable of forming the optical interconnection weight matrix, which may have large dynamic range and can be quickly erased and re writen by optical methods.
出处
《光学学报》
EI
CAS
CSCD
北大核心
1997年第6期766-771,共6页
Acta Optica Sinica
基金
国家自然科学基金
关键词
电子俘获
IPA
神经网络模型
互联权重矩阵
electron trapping materials, IPA neural network, interconnection weight matrix.
