基于运算放大器的LIF神经元和RRAM突触阵列的脉冲神经网络电路

IF 3.2 3区计算机科学 Q2 ENGINEERING, ELECTRICAL & ELECTRONIC

Aeu-International Journal of Electronics and Communications Pub Date : 2025-08-20 DOI:10.1016/j.aeue.2025.156004

Eashwar M.V. , Nivetha T. , Bindu B. , Noor Ain Kamsani

{"title":"基于运算放大器的LIF神经元和RRAM突触阵列的脉冲神经网络电路","authors":"Eashwar M.V. , Nivetha T. , Bindu B. , Noor Ain Kamsani","doi":"10.1016/j.aeue.2025.156004","DOIUrl":null,"url":null,"abstract":"<div><div>Spiking Neural Networks (SNNs) have become crucial in neuromorphic computing for efficiently processing the vast amounts of digital data generated in the technology-driven world. The resistive RAM (RRAM) based SNNs offer superior energy efficiency, high-speed processing, parallelism, and scalability for neuromorphic computing applications. In this article, an SNN circuit with a 1T-1R RRAM synaptic array along with op-amp and 555 timer-based leaky integrate-and-fire (LIF) neuron is implemented to use for pattern recognition. The input pre-spikes from the pattern are applied to the RRAM synaptic array, which exhibits synaptic plasticity. The LIF neuron processes the synaptic array output to produce post-spikes, which modify the conductance of the RRAM synaptic array based on the spike-timing-dependent plasticity (STDP) mechanism. The unique output spikes obtained for different characters can be used for pattern recognition of the characters.</div></div>","PeriodicalId":50844,"journal":{"name":"Aeu-International Journal of Electronics and Communications","volume":"201 ","pages":"Article 156004"},"PeriodicalIF":3.2000,"publicationDate":"2025-08-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Spiking neural network circuit with op-amp based LIF neuron and RRAM synaptic array\",\"authors\":\"Eashwar M.V. , Nivetha T. , Bindu B. , Noor Ain Kamsani\",\"doi\":\"10.1016/j.aeue.2025.156004\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<div><div>Spiking Neural Networks (SNNs) have become crucial in neuromorphic computing for efficiently processing the vast amounts of digital data generated in the technology-driven world. The resistive RAM (RRAM) based SNNs offer superior energy efficiency, high-speed processing, parallelism, and scalability for neuromorphic computing applications. In this article, an SNN circuit with a 1T-1R RRAM synaptic array along with op-amp and 555 timer-based leaky integrate-and-fire (LIF) neuron is implemented to use for pattern recognition. The input pre-spikes from the pattern are applied to the RRAM synaptic array, which exhibits synaptic plasticity. The LIF neuron processes the synaptic array output to produce post-spikes, which modify the conductance of the RRAM synaptic array based on the spike-timing-dependent plasticity (STDP) mechanism. The unique output spikes obtained for different characters can be used for pattern recognition of the characters.</div></div>\",\"PeriodicalId\":50844,\"journal\":{\"name\":\"Aeu-International Journal of Electronics and Communications\",\"volume\":\"201 \",\"pages\":\"Article 156004\"},\"PeriodicalIF\":3.2000,\"publicationDate\":\"2025-08-20\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Aeu-International Journal of Electronics and Communications\",\"FirstCategoryId\":\"94\",\"ListUrlMain\":\"https://www.sciencedirect.com/science/article/pii/S1434841125003450\",\"RegionNum\":3,\"RegionCategory\":\"计算机科学\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q2\",\"JCRName\":\"ENGINEERING, ELECTRICAL & ELECTRONIC\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Aeu-International Journal of Electronics and Communications","FirstCategoryId":"94","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S1434841125003450","RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"ENGINEERING, ELECTRICAL & ELECTRONIC","Score":null,"Total":0}

引用次数: 0

摘要

脉冲神经网络（snn）在神经形态计算中已经变得至关重要，它可以有效地处理技术驱动世界中产生的大量数字数据。基于电阻性RAM （RRAM）的snn为神经形态计算应用提供了卓越的能效、高速处理、并行性和可扩展性。在本文中，实现了一个SNN电路，该电路具有1T-1R RRAM突触阵列以及运算放大器和基于555定时器的泄漏集成与发射（LIF）神经元，用于模式识别。来自该模式的输入预尖峰被应用于RRAM突触阵列，其表现出突触可塑性。LIF神经元处理突触阵列输出产生后峰值，后峰值基于峰值时间依赖的可塑性（STDP）机制改变RRAM突触阵列的电导。对不同字符获得的唯一输出尖峰可以用于字符的模式识别。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

查看原文本刊更多论文

Spiking neural network circuit with op-amp based LIF neuron and RRAM synaptic array

Spiking Neural Networks (SNNs) have become crucial in neuromorphic computing for efficiently processing the vast amounts of digital data generated in the technology-driven world. The resistive RAM (RRAM) based SNNs offer superior energy efficiency, high-speed processing, parallelism, and scalability for neuromorphic computing applications. In this article, an SNN circuit with a 1T-1R RRAM synaptic array along with op-amp and 555 timer-based leaky integrate-and-fire (LIF) neuron is implemented to use for pattern recognition. The input pre-spikes from the pattern are applied to the RRAM synaptic array, which exhibits synaptic plasticity. The LIF neuron processes the synaptic array output to produce post-spikes, which modify the conductance of the RRAM synaptic array based on the spike-timing-dependent plasticity (STDP) mechanism. The unique output spikes obtained for different characters can be used for pattern recognition of the characters.

求助全文

通过发布文献求助，成功后即可免费获取论文全文。去求助

来源期刊

Aeu-International Journal of Electronics and Communications 工程技术-电信学

CiteScore

6.90

自引率

18.80%

发文量

292

审稿时长

4.9 months

期刊介绍： AEÜ is an international scientific journal which publishes both original works and invited tutorials. The journal''s scope covers all aspects of theory and design of circuits, systems and devices for electronics, signal processing, and communication, including: signal and system theory, digital signal processing network theory and circuit design information theory, communication theory and techniques, modulation, source and channel coding switching theory and techniques, communication protocols optical communications microwave theory and techniques, radar, sonar antennas, wave propagation AEÜ publishes full papers and letters with very short turn around time but a high standard review process. Review cycles are typically finished within twelve weeks by application of modern electronic communication facilities.