Photonic integrated neuro-synaptic core for convolutional spiking neural network

IF 15.3 1区物理与天体物理 Q1 OPTICS

Opto-Electronic Advances Pub Date : 2023-01-01 DOI:10.29026/oea.2023.230140

Shuiying Xiang, Yuechun Shi, Yahui Zhang, Xingxing Guo, Ling Zheng, Yanan Han, Yuna Zhang, Ziwei Song, Dianzhuang Zheng, Tao Zhang, Hailing Wang, Xiaojun Zhu, Xiangfei Chen, Min Qiu, Yichen Shen, Wanhua Zheng, Yue Hao

{"title":"Photonic integrated neuro-synaptic core for convolutional spiking neural network","authors":"Shuiying Xiang, Yuechun Shi, Yahui Zhang, Xingxing Guo, Ling Zheng, Yanan Han, Yuna Zhang, Ziwei Song, Dianzhuang Zheng, Tao Zhang, Hailing Wang, Xiaojun Zhu, Xiangfei Chen, Min Qiu, Yichen Shen, Wanhua Zheng, Yue Hao","doi":"10.29026/oea.2023.230140","DOIUrl":null,"url":null,"abstract":"Neuromorphic photonic computing has emerged as a competitive computing paradigm to overcome the bottlenecks of the von-Neumann architecture. Linear weighting and nonlinear spike activation are two fundamental functions of a photonic spiking neural network (PSNN). However, they are separately implemented with different photonic materials and devices, hindering the large-scale integration of PSNN. Here, we propose, fabricate and experimentally demonstrate a photonic neuro-synaptic chip enabling the simultaneous implementation of linear weighting and nonlinear spike activation based on a distributed feedback (DFB) laser with a saturable absorber (DFB-SA). A prototypical system is experimentally constructed to demonstrate the parallel weighted function and nonlinear spike activation. Furthermore, a four-channel DFB-SA laser array is fabricated for realizing matrix convolution of a spiking convolutional neural network, achieving a recognition accuracy of 87% for the MNIST dataset. The fabricated neuro-synaptic chip offers a fundamental building block to construct the large-scale integrated PSNN chip.","PeriodicalId":19611,"journal":{"name":"Opto-Electronic Advances","volume":"49 1","pages":"0"},"PeriodicalIF":15.3000,"publicationDate":"2023-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Opto-Electronic Advances","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.29026/oea.2023.230140","RegionNum":1,"RegionCategory":"物理与天体物理","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"OPTICS","Score":null,"Total":0}

引用次数: 0

Abstract

Neuromorphic photonic computing has emerged as a competitive computing paradigm to overcome the bottlenecks of the von-Neumann architecture. Linear weighting and nonlinear spike activation are two fundamental functions of a photonic spiking neural network (PSNN). However, they are separately implemented with different photonic materials and devices, hindering the large-scale integration of PSNN. Here, we propose, fabricate and experimentally demonstrate a photonic neuro-synaptic chip enabling the simultaneous implementation of linear weighting and nonlinear spike activation based on a distributed feedback (DFB) laser with a saturable absorber (DFB-SA). A prototypical system is experimentally constructed to demonstrate the parallel weighted function and nonlinear spike activation. Furthermore, a four-channel DFB-SA laser array is fabricated for realizing matrix convolution of a spiking convolutional neural network, achieving a recognition accuracy of 87% for the MNIST dataset. The fabricated neuro-synaptic chip offers a fundamental building block to construct the large-scale integrated PSNN chip.

查看原文本刊更多论文

卷积脉冲神经网络的光子集成神经突触核

神经形态光子计算已经成为一种有竞争力的计算范式，以克服冯-诺伊曼架构的瓶颈。线性加权和非线性尖峰激活是光子尖峰神经网络的两个基本功能。然而，它们是用不同的光子材料和器件单独实现的，阻碍了PSNN的大规模集成。在这里，我们提出，制造和实验证明了一个光子神经突触芯片，能够同时实现线性加权和非线性尖峰激活基于分布式反馈(DFB)激光器与饱和吸收器(DFB- sa)。实验建立了一个原型系统，验证了并行加权函数和非线性尖峰激活。此外，制作了一个四通道DFB-SA激光阵列，用于实现尖峰卷积神经网络的矩阵卷积，对MNIST数据集的识别精度达到87%。所制备的神经突触芯片为大规模集成PSNN芯片的构建提供了基础模块。

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

Opto-Electronic Advances OPTICS-

CiteScore

19.30

自引率

7.10%

发文量

128

期刊介绍： Opto-Electronic Advances (OEA) is a distinguished scientific journal that has made significant strides since its inception in March 2018. Here's a collated summary of its key features and accomplishments: Impact Factor and Ranking: OEA boasts an impressive Impact Factor of 14.1, which positions it within the Q1 quartiles of the Optics category. This high ranking indicates that the journal is among the top 25% of its field in terms of citation impact. Open Access and Peer Review: As an open access journal, OEA ensures that research findings are freely available to the global scientific community, promoting wider dissemination and collaboration. It upholds rigorous academic standards through a peer review process, ensuring the quality and integrity of the published research. Database Indexing: OEA's content is indexed in several prestigious databases, including the Science Citation Index (SCI), Engineering Index (EI), Scopus, Chemical Abstracts (CA), and the Index to Chinese Periodical Articles (ICI). This broad indexing facilitates easy access to the journal's articles by researchers worldwide. Scope and Purpose: OEA is committed to serving as a platform for the exchange of knowledge through the publication of high-quality empirical and theoretical research papers. It covers a wide range of topics within the broad area of optics, photonics, and optoelectronics, catering to researchers, academicians, professionals, practitioners, and students alike.