Analog Spiking Neural Network Synthesis for the MNIST

Q4 Engineering
Thomas Soupizet, Zalfa Jouni, Siqi Wang, A. Benlarbi-Delai, Pietro M. Ferreira
{"title":"Analog Spiking Neural Network Synthesis for the MNIST","authors":"Thomas Soupizet, Zalfa Jouni, Siqi Wang, A. Benlarbi-Delai, Pietro M. Ferreira","doi":"10.29292/jics.v18i1.663","DOIUrl":null,"url":null,"abstract":"Different from classical artificial neural network which processes digital data, the spiking neural network (SNN) processes spike trains. Indeed, its event-driven property helps to capture the rich dynamics the neurons have within the brain, and the sparsity of collected spikes helps reducing computational power. Novel synthesis framework is proposed and an algorithm is detailed to guide designers into deep learning and energy-efficient analog SNN using MNIST. An analog SNN composed of 86 electronic neurons (eNeuron) and 1238 synapses interacting through two hidden layers is illustrated. Three different models of eNeurons implementations are tested, being (Leaky) Integrate-and-Fire (LIF), Morris Lecar (ML) simplified (simp.) and biomimetic (bio.). The proposed SNN, coupling deep learning and ultra-low power, is trained using a common machine learning system (Tensor- Flow) for the MNIST. LIF eNeurons implementations present some limitations and weakness in terms of dynamic range. Both ML eNeurons achieve robust accuracy which is approximately of 0.82.","PeriodicalId":39974,"journal":{"name":"Journal of Integrated Circuits and Systems","volume":" ","pages":""},"PeriodicalIF":0.0000,"publicationDate":"2023-05-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"2","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Journal of Integrated Circuits and Systems","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.29292/jics.v18i1.663","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q4","JCRName":"Engineering","Score":null,"Total":0}
引用次数: 2

Abstract

Different from classical artificial neural network which processes digital data, the spiking neural network (SNN) processes spike trains. Indeed, its event-driven property helps to capture the rich dynamics the neurons have within the brain, and the sparsity of collected spikes helps reducing computational power. Novel synthesis framework is proposed and an algorithm is detailed to guide designers into deep learning and energy-efficient analog SNN using MNIST. An analog SNN composed of 86 electronic neurons (eNeuron) and 1238 synapses interacting through two hidden layers is illustrated. Three different models of eNeurons implementations are tested, being (Leaky) Integrate-and-Fire (LIF), Morris Lecar (ML) simplified (simp.) and biomimetic (bio.). The proposed SNN, coupling deep learning and ultra-low power, is trained using a common machine learning system (Tensor- Flow) for the MNIST. LIF eNeurons implementations present some limitations and weakness in terms of dynamic range. Both ML eNeurons achieve robust accuracy which is approximately of 0.82.
MNIST的模拟Spiking神经网络综合
与处理数字数据的经典人工神经网络不同,尖峰神经网络处理尖峰序列。事实上,它的事件驱动特性有助于捕捉神经元在大脑中的丰富动态,而收集到的尖峰的稀疏性有助于降低计算能力。提出了一种新的综合框架,并详细介绍了一种算法,以指导设计者使用MNIST进行深度学习和节能模拟SNN。图示了由86个电子神经元(eNeuron)和1238个突触通过两个隐藏层相互作用组成的模拟SNN。测试了三种不同的eNeurons实现模型,即(Leaky)Integration and Fire(LIF)、Morris Lecar(ML)simp.和仿生(bio.)。所提出的SNN耦合深度学习和超低功率,使用MNIST的通用机器学习系统(Tensor-Flow)进行训练。LIF eNeurons的实现在动态范围方面存在一些局限性和弱点。两个ML eNeuron都实现了大约为0.82的鲁棒精度。
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 求助全文
来源期刊
Journal of Integrated Circuits and Systems
Journal of Integrated Circuits and Systems Engineering-Electrical and Electronic Engineering
CiteScore
0.90
自引率
0.00%
发文量
39
期刊介绍: This journal will present state-of-art papers on Integrated Circuits and Systems. It is an effort of both Brazilian Microelectronics Society - SBMicro and Brazilian Computer Society - SBC to create a new scientific journal covering Process and Materials, Device and Characterization, Design, Test and CAD of Integrated Circuits and Systems. The Journal of Integrated Circuits and Systems is published through Special Issues on subjects to be defined by the Editorial Board. Special issues will publish selected papers from both Brazilian Societies annual conferences, SBCCI - Symposium on Integrated Circuits and Systems and SBMicro - Symposium on Microelectronics Technology and Devices.
×
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
确定
请完成安全验证×
copy
已复制链接
快去分享给好友吧!
我知道了
右上角分享
点击右上角分享
0
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 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号:481959085
Book学术官方微信