An Optical Accelerator for Deep Neural Network Based on Integrated Nanophotonics

被引:0
|
作者
Shiomi, Jun [1 ]
Ishihara, Tohru [2 ]
Onodera, Hidetoshi [1 ]
Shinya, Akihiko [3 ]
Notomi, Masaya [3 ]
机构
[1] Kyoto Univ, Grad Sch Informat, Sakyo Ku, Kyoto 6068501, Japan
[2] Nagoya Univ, Grad Sch Informat, Chikusa Ku, Nagoya, Aichi 4648603, Japan
[3] NTT Basic Res Labs, NTT Nanophoton Ctr, 3-1 Morinosato Wakamiya, Atsugi, Kanagawa 2430198, Japan
关键词
optical computing; deep neural network; integrated nanophotonics;
D O I
10.1109/ICRC2020.2020.00017
中图分类号
TP301 [理论、方法];
学科分类号
081202 ;
摘要
The emergence of nanophotonic devices has enabled to design light-speed on-chip optical circuits with extremely low latency. This paper proposes an optical implementation of scalable Deep Neural Networks (DNNs) enabling light-speed inference. The key issue in optical neural networks is the scalability limited by area, power and the number of available wavelengths. Due to the scalability, it is thus difficult to design an all-optical hardware accelerator for a large-scale DNN. To solve this problem, this paper firstly proposes an optical Vector Matrix Multiplier (VMM) structure operating with a low latency. The multipliers in a VMM are highly parallelized based on the Wavelength Division Multiplexing (WDM) technique, which reduces the area overhead without sacrificing the ultra-high speed nature. This paper then proposes the electrical digital interfaces for storing and handling intermediate VMM data without sacrificing the ultra-high speed nature, which enables to reuse the VMM multiple times with a low latency.
引用
收藏
页码:95 / 101
页数:7
相关论文
共 50 条
  • [1] Inverse design of an integrated-nanophotonics optical neural network
    Qu, Yurui
    Zhu, Huanzheng
    Shen, Yichen
    Zhang, Jin
    Tao, Chenning
    Ghosh, Pintu
    Qiu, Min
    [J]. SCIENCE BULLETIN, 2020, 65 (14) : 1177 - 1183
  • [2] Deep Neural Network Accelerator based on FPGA
    Thang Viet Huynh
    [J]. 2017 4TH NAFOSTED CONFERENCE ON INFORMATION AND COMPUTER SCIENCE (NICS), 2017, : 254 - 257
  • [3] Activation in Network for NoC-based Deep Neural Network Accelerator
    Zhu, Wenyao
    Chen, Yizhi
    Lu, Zhonghai
    [J]. 2024 INTERNATIONAL VLSI SYMPOSIUM ON TECHNOLOGY, SYSTEMS AND APPLICATIONS, VLSI TSA, 2024,
  • [4] Optical Neural Network in Free-Space and Nanophotonics
    Sun, Zhenlin
    Yu, Miao
    Song, Zhengxun
    Liu, Weiwen
    Xing, Gangyao
    Zhou, Muhan
    [J]. IEEE ACCESS, 2023, 11 : 88656 - 88669
  • [5] DNNARA: A Deep Neural Network Accelerator using Residue Arithmetic and Integrated Photonics
    Peng, Jiaxin
    Alkabani, Yousra
    Sun, Shuai
    Sorger, Volker J.
    El-Ghazawi, Tarek
    [J]. PROCEEDINGS OF THE 49TH INTERNATIONAL CONFERENCE ON PARALLEL PROCESSING, ICPP 2020, 2020,
  • [6] A Deep Neural Network Accelerator Based on Tiled RRAM Architecture
    Wang, Qiwen
    Wang, Xinxin
    Lee, Seung Hwan
    Meng, Fan-Hsuan
    Lu, Wei D.
    [J]. 2019 IEEE INTERNATIONAL ELECTRON DEVICES MEETING (IEDM), 2019,
  • [7] Deep neural network-based phase calibration in integrated optical phased arrays
    Jae-Yong Kim
    Junhyeong Kim
    Jinhyeong Yoon
    Seokjin Hong
    Berkay Neseli
    Namhyun Kwon
    Jong-Bum You
    Hyeonho Yoon
    Hyo-Hoon Park
    Hamza Kurt
    [J]. Scientific Reports, 13 (1)
  • [8] Deep neural network-based phase calibration in integrated optical phased arrays
    Kim, Jae-Yong
    Kim, Junhyeong
    Yoon, Jinhyeong
    Hong, Seokjin
    Neseli, Berkay
    Kwon, Namhyun
    You, Jong-Bum
    Yoon, Hyeonho
    Park, Hyo-Hoon
    Kurt, Hamza
    [J]. SCIENTIFIC REPORTS, 2023, 13 (01):
  • [9] Integrated WDM-compatible optical mode division multiplexing neural network accelerator
    Yin, Ruoyu
    Xiao, Huifu
    Jiang, Yongheng
    Han, Xu
    Zhang, Pu
    Chen, Li
    Zhou, Xudong
    Yuan, Mingrui
    Ren, Guanghui
    Mitchell, Arnan
    Tian, Yonghui
    [J]. OPTICA, 2023, 10 (12): : 1709 - 1718
  • [10] Integrated Nanophotonics for Optical Computation
    Notomi, M.
    [J]. 2017 CONFERENCE ON LASERS AND ELECTRO-OPTICS (CLEO), 2017,