• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文会议>International Joint Conference on Neural Networks >Development of resistive memories based on silver doped graphene oxide for neuron simulation
【24h】

Development of resistive memories based on silver doped graphene oxide for neuron simulation

机译：基于掺杂银的氧化石墨烯的电阻存储器的神经元仿真开发

获取原文
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

Resistive memory (ReRAM) can be approximately one hundred times faster than current Flash technologies. This device can still be more economical in power consumption. The key element that makes ReRAM memories so attractive is a component called memristor. In this work, device fabrication used doped graphene oxide with silver (GO + 0.1% Ag). ITON (Indium Tin oxynitride) and aluminum were used as contacts. Thin films are obtained by the dip coating technique. The mechanism of the resistive switching effect in doped graphene oxide thin film based devices has been investigated by macroscopic current-voltage (IxV) measurements. Device was tested individually; its electrical characteristics and its behavior were verified in case of voltage variations in order to establish the operating limits for set and reset. From the experimental results, a simulation of neural networks was made using data obtained. The neural network model chosen was the so-called perceptron. This is the simplest form, imitating a single neuron that makes the weighted sum of stimuli in its dendrinos and applies an activation function to decide if it is the case of sending or not a signal in its axon, exciting or inhibiting other neurons. Only two synapses were used, simulating only two characteristics to maintain simplicity both in the example and the presentation of the data. However the example can be expanded to more synapses, inserting more bridges of memristors or even evolving into a multilayer perceptron.
机译：电阻存储器（ReRAM）的速度大约是当前闪存技术的一百倍。该设备在功耗上仍然可以更加经济。使ReRAM存储器如此吸引人的关键因素是一个称为忆阻器的组件。在这项工作中，器件制造使用了掺杂了银的氧化石墨烯（GO + 0.1％Ag）。 ITON（氧氮化铟锡）和铝用作触点。通过浸涂技术获得薄膜。已通过宏观电流-电压（IxV）测量研究了掺杂氧化石墨烯薄膜基器件中的电阻切换效应的机理。设备已单独测试;在电压变化的情况下，对其电气特性和性能进行了验证，以便确定设置和复位的操作极限。根据实验结果，使用获得的数据对神经网络进行了仿真。选择的神经网络模型是所谓的感知器。这是最简单的形式，它模仿单个神经元，该神经元在其树突中产生加权的总和，并应用激活函数来确定是否是在其轴突中发送信号，激发或抑制其他神经元的情况。仅使用了两个突触，仅在示例和数据表示中模拟了两个特征以保持简单性。但是，该示例可以扩展为更多的突触，插入更多的忆阻器桥，甚至演变为多层感知器。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号