离散Hopfield神经网络和联想记忆的开关电流技术实现

Implementation of Discrete Hopfield Neural Network and Associative Memory Using Switched Current Techniques

研究了离散Ｈｏｐｆｉｅｌｄ神经网络（ＤＨＮＮ）和联想记忆神经网络的开关电流技术实现。利用多权输入跨导、开关电流延迟器（ＳＩＤ）和可编程电流比较器（ＰＣＣ）实现了离散Ｈｏｐｆｉｅｌｄ神经网络，并提出了利用离散Ｈｏｐｆｉｅｌｄ神经网络实现自联想记忆时相应的开关电流电路。所提出的开关电流神经网络适宜于超大规模集成，能在低电压（如３．３Ｖ）下工作。

A systematic approach is developed for the implementation of discrete Hopfield neural network(DHNN)and associative memory(AM) neural network using switched current(SI)integrated circuit techniques Three neuron cells,which are multiple input transconductor,switched current delay(SID)and programmable current comparator(PCC)used for the implementation of DHNN,are introduced A switched current AM circuit architecture is also presented The proposed SI neural circuits is capable of operating at power supply down to 3 3 V,and it is suitable for VLSI process