ScholarMate
English简体中文繁體中文
客服热线:400-1616-289
登录注册
期刊论文

Fully forward mode training for optical neural networks

Xue, Zhiwei; Zhou, Tiankuang; Xu, Zhihao; Yu, Shaoliang; Dai, Qionghai*; Fang, Lu*清华大学
Nature2024-8-8DOI: 10.1038/s41586-024-07687-4
2被引1阅读
引用收藏
分享

摘要

Optical computing promises to improve the speed and energy efficiency of machine learning applications1-6. However, current approaches to efficiently train these models are limited by in silico emulation on digital computers. Here we develop a method called fully forward mode (FFM) learning, which implements the compute-intensive training process on the physical system. The majority of the machine learning operations are thus efficiently conducted in parallel on site, alleviating numerical modelling constraints. In free-space and integrated photonics, we experimentally demonstrate optical systems with state-of-the-art performances for a given network size. FFM learning shows training the deepest optical neural networks with millions of parameters achieves accuracy equivalent to the ideal model. It supports all-optical focusing through scattering media with a resolution of the diffraction limit; it can also image in parallel the objects hidden outside the direct line of sight at over a kilohertz frame rate and can conduct all-optical processing with light intensity as weak as subphoton per pixel (5.40 x 1018- operations-per-second-per-watt energy efficiency) at room temperature. Furthermore, we prove that FFM learning can automatically search non-Hermitian exceptional points without an analytical model. FFM learning not only facilitates orders-of-magnitude-faster learning processes, but can also advance applied and theoretical fields such as deep neural networks, ultrasensitive perception and topological photonics. We present fully forward mode learning, which conducts machine learning operations on site, leading to faster learning and promoting advancement in numerous fields.

学科与关键词

关键词
INVERSE DESIGNARTIFICIAL-INTELLIGENCEBACKPROPAGATIONTIME
相关
加载中…
  • Role of all-optical neural networks
    Matuszewski, M.*; Prystupiuk, A.; Opala, A.Physical Review Applied, 2024, 21(1): 014028.被引 6
  • At the intersection of optics and deep learning: statistical inference, computing, and inverse design
    Mengu, Deniz; Rahman, Md Sadman Sakib; Luo, Yi; Li, Jingxi; Kulce, Onur; Ozcan, Aydogan*Advances in Optics and Photonics, 2022, 14(2): 209-290.被引 34
  • An optical neural network using less than 1 photon per multiplication
    Wang, Tianyu*; Ma, Shi-Yuan; Wright, Logan G.; Onodera, Tatsuhiro; Richard, Brian C.; McMahon, Peter L.*Nature Communications, 2022, 13(1): 123.被引 122
查看全部 →

发表信息

论文状态
公开发表
期刊
Nature
发表日期
2024-08-08
卷期
第 632 卷, 第 8024 期
DOI
10.1038/s41586-024-07687-4
语言
ENGLISH
成果收录
SCIE
资助标注
National Science and Technology Major Project [2021ZD0109901]; National Natural Science Foundation of China (NSFC) [62125106, 62088102]; China Association for Science and Technology (CAST) [2023QNRC001]; China Postdoctoral Science Foundation (CPSF) [GZB20230372]

作者

  • X

    Xue, Zhiwei (第一作者)

    tsinghua univ

  • Z

    Zhou, Tiankuang

    tsinghua univ

查看全部 6 位作者收起作者
  • X

    Xu, Zhihao

    tsinghua univ

  • Y

    Yu, Shaoliang

  • D

    Dai, Qionghai (通讯)

    tsinghua univ

  • F

    Fang, Lu (通讯)

    tsinghua univ

登录解锁更多功能

  • 收藏这篇,随时在“我的 · 收藏”找回
  • 认领你的成果,生成推广素材
  • 关注作者获取最新成果
  • 加入项目组、协同写作
注册 / 登录

产品服务

  • 科研之友
  • 创新城
  • 科创云

服务支持

  • 帮助中心
  • 隐私政策
  • 服务条款

联系方式

在线客服:【】
客服热线:400-1616-289
电子邮箱:support@scholarmate.com

关注或下载科研之友

微信二维码
微信公众号
客户端下载二维码
下载客户端
科研成果科研人员科研机构科研动态爱瑞思软件

©2026 深圳市科研之友网络服务有限公司

公安备案图标粤公网安备 44030502000213
粤ICP备 16046710 号粤B2-20110417