4.6 Article

A Stacked Memristive Device Enabling Both Analog and Threshold Switching Behaviors for Artificial Leaky Integrate and Fire Neuron

期刊

IEEE ELECTRON DEVICE LETTERS
卷 43, 期 9, 页码 1436-1439

出版社

IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
DOI: 10.1109/LED.2022.3188786

关键词

Neurons; Biology; Memristors; Switches; Threshold voltage; Nanobioscience; Immune system; Artificial neuron; memristive device; threshold switching; analog switching

资金

  1. Ministry of Science and Technology of China [2018YFE0118300, 2019YFB2205100]
  2. National Science Fund for Distinguished Young Scholars [52025022]
  3. National Nature Science Foundation of China [U19A2091, 62004016, 51732003, 20210509045RQ, 51872043, 51842201, 52072065, 61574031]
  4. 111 Project [B13013]

向作者/读者索取更多资源

This paper introduces a stacked memristor-based LIF neuron that can accomplish integration and firing functions and successfully mimics key functions of biological neurons. The fabricated neurons have great potential for constructing high-density spiking neural networks for neuromorphic computing.
Leaky integrate and fire (LIF) neurons are critical units for constructing a spiking neural network, in which neurons communicate with each other using spikes via synapses. Memristors, due to its specific nonlinear characteristics, are frequently introduced to emulate partial functions of LIF neurons for simplifying the circuit complexity, either the integration process or the fire action. Usually, a relatively complicated peripheral circuit needs to be engineered to assist the memristive device for complete emulation for biological neurons, which certainly would hinder the integration potential. Herein, we fabricated a stacked memristive device possessing both analog and threshold switching behaviors for constructing an artificial LIF neuron. Thus, the integration and fire functions were both accomplished within this single nanoscale device. In addition, the key neuronic functional of a biological neuron, including all-or-nothing spiking, threshold spiking, a refractory period, and strength-modulated frequency response were all successfully mimicked. The results demonstrate that the fabricated stacked memristor-based LIF neurons have great potential to construct high-density spiking neural network for neuromorphic computing.

作者

我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。

评论

主要评分

4.6
评分不足

次要评分

新颖性
-
重要性
-
科学严谨性
-
评价这篇论文

推荐

暂无数据
暂无数据