An energy-efficient tunable threshold spiking neuron with excitatory and inhibitory function

IF 1.6 4区 工程技术 Q3 ENGINEERING, ELECTRICAL & ELECTRONIC International Journal of Numerical Modelling-Electronic Networks Devices and Fields Pub Date : 2024-03-15 DOI:10.1002/jnm.3227
Mudasir A. Khanday, Farooq A. Khanday
{"title":"An energy-efficient tunable threshold spiking neuron with excitatory and inhibitory function","authors":"Mudasir A. Khanday,&nbsp;Farooq A. Khanday","doi":"10.1002/jnm.3227","DOIUrl":null,"url":null,"abstract":"<p>In this work, a complementary metal-oxide-semiconductor (CMOS) based leaky-integrate and fire neuron has been proposed and investigated for neuromorphic applications. The neuron has been designed in Cadence Virtuoso and validated experimentally. It has been observed that the neuron consumes a maximum energy of 68.87 fJ/spike. The response of the neuron to excitatory as well as inhibitory inputs has been studied. To verify the applicability, the proposed neuron has been explored for reconfigurable threshold logic to implement various linearly separable Boolean functions including OR, AND, NOT, NOR, and NAND. Moreover, the threshold tunability of the neuron has also been verified and this property has been exploited to design threshold-controlled logic gates. Instead of adjusting the weights of the applied inputs, the functionality of such gates can be controlled by changing the threshold of the neuron, simplifying the synaptic architecture of a neural network. Finally, a multilayer network has been designed and the recognition ability of the proposed network for MNIST handwritten digits has been verified with an accuracy of 96.93%.</p>","PeriodicalId":50300,"journal":{"name":"International Journal of Numerical Modelling-Electronic Networks Devices and Fields","volume":"37 2","pages":""},"PeriodicalIF":1.6000,"publicationDate":"2024-03-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"International Journal of Numerical Modelling-Electronic Networks Devices and Fields","FirstCategoryId":"5","ListUrlMain":"https://onlinelibrary.wiley.com/doi/10.1002/jnm.3227","RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q3","JCRName":"ENGINEERING, ELECTRICAL & ELECTRONIC","Score":null,"Total":0}
引用次数: 0

Abstract

In this work, a complementary metal-oxide-semiconductor (CMOS) based leaky-integrate and fire neuron has been proposed and investigated for neuromorphic applications. The neuron has been designed in Cadence Virtuoso and validated experimentally. It has been observed that the neuron consumes a maximum energy of 68.87 fJ/spike. The response of the neuron to excitatory as well as inhibitory inputs has been studied. To verify the applicability, the proposed neuron has been explored for reconfigurable threshold logic to implement various linearly separable Boolean functions including OR, AND, NOT, NOR, and NAND. Moreover, the threshold tunability of the neuron has also been verified and this property has been exploited to design threshold-controlled logic gates. Instead of adjusting the weights of the applied inputs, the functionality of such gates can be controlled by changing the threshold of the neuron, simplifying the synaptic architecture of a neural network. Finally, a multilayer network has been designed and the recognition ability of the proposed network for MNIST handwritten digits has been verified with an accuracy of 96.93%.

查看原文
分享 分享
微信好友 朋友圈 QQ好友 复制链接
本刊更多论文
具有兴奋和抑制功能的高能效可调阈值尖峰神经元
在这项工作中,我们提出了一种基于互补金属氧化物半导体(CMOS)的漏积分和火神经元,并针对神经形态应用进行了研究。该神经元是在 Cadence Virtuoso 中设计的,并经过了实验验证。据观察,该神经元消耗的最大能量为 68.87 fJ/spike。研究了神经元对兴奋和抑制输入的响应。为了验证神经元的适用性,研究人员探索了所提出的神经元的可重构阈值逻辑,以实现各种线性可分离布尔函数,包括 OR、AND、NOT、NOR 和 NAND。此外,还验证了神经元的阈值可调性,并利用这一特性设计了阈值控制逻辑门。这种逻辑门的功能可以通过改变神经元的阈值来控制,而不是调整应用输入的权重,从而简化了神经网络的突触结构。最后,我们设计了一个多层网络,并验证了该网络对 MNIST 手写数字的识别能力,准确率高达 96.93%。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
求助全文
约1分钟内获得全文 去求助
来源期刊
CiteScore
4.60
自引率
6.20%
发文量
101
审稿时长
>12 weeks
期刊介绍: Prediction through modelling forms the basis of engineering design. The computational power at the fingertips of the professional engineer is increasing enormously and techniques for computer simulation are changing rapidly. Engineers need models which relate to their design area and which are adaptable to new design concepts. They also need efficient and friendly ways of presenting, viewing and transmitting the data associated with their models. The International Journal of Numerical Modelling: Electronic Networks, Devices and Fields provides a communication vehicle for numerical modelling methods and data preparation methods associated with electrical and electronic circuits and fields. It concentrates on numerical modelling rather than abstract numerical mathematics. Contributions on numerical modelling will cover the entire subject of electrical and electronic engineering. They will range from electrical distribution networks to integrated circuits on VLSI design, and from static electric and magnetic fields through microwaves to optical design. They will also include the use of electrical networks as a modelling medium.
期刊最新文献
Accelerated Characteristic Mode Calculation for PEC Objects Using ACA-QR-SVD Algorithm Enhanced Performance of Dual Material Double Gate Negative Capacitance Tunnel Field Effect Transistor (DMDG-NC-TFET) via HZO Ferroelectric Integration for Improved Drain Current and Subthreshold Swing EEG Emotion Recognition Based on GADF and AMB-CNN Model A High Power Density Ku-Band GaN Power Amplifier Based on Device-Level Thermal Analysis One-Step Leapfrog 3D Split-Field FDTD Method for Periodic Structures at Oblique Incidence
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
现在去查看 取消
×
提示
确定
0
微信
客服QQ
Book学术公众号 扫码关注我们
反馈
×
意见反馈
请填写您的意见或建议
请填写您的手机或邮箱
已复制链接
已复制链接
快去分享给好友吧!
我知道了
×
扫码分享
扫码分享
Book学术官方微信
Book学术文献互助
Book学术文献互助群
群 号:481959085
Book学术
文献互助 智能选刊 最新文献 互助须知 联系我们:info@booksci.cn
Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。
Copyright © 2023 Book学术 All rights reserved.
ghs 京公网安备 11010802042870号 京ICP备2023020795号-1