.png){kind=icon}
返回
Single-shot optical neural network
.png){kind=icon}
OA
AI
DOI：10.1126/sciadv.adg7904.png)
.png){kind=icon}
.png){kind=icon}
分享.png){kind=icon}
收藏
摘要
En 中文
Analog optical and electronic hardware has emerged as a promising alternative to digital electronics to improve the efficiency of deep neural networks (DNNs). However, previous work has been limited in scalability (input vector length K approximate to 100 elements) or has required nonstandard DNN models and retraining, hindering widespread adoption. Here, we present an analog, CMOS-compatible DNN processor that uses free-space optics to reconfigurably distribute an input vector and optoelectronics for static, updatable weighting and the nonlinearity-with K approximate to 1000 and beyond. We demonstrate single-shot-per-layer classification of the MNIST, Fashion-MNIST, and QuickDraw datasets with standard fully connected DNNs, achieving respective accuracies of 95.6, 83.3, and 79.0% without preprocessing or retraining. We also experimentally determine the fundamental upper bound on throughput (similar to 0.9 exaMAC/s), set by the maximum optical bandwidth before substantial increase in error. Our combination of wide spectral and spatial bandwidths enables highly efficient computing for next-generation DNNs.
Keyword:
PARALLEL
GENERATION
PHOTON
AI总结
对已上传原文的论文进行重点信息的提取，主要内容包括：简要概述、研究摘要、背景介绍、关键亮点、图文解析、展望与总结。
期刊
IF:
12.5
论文数:
2.1W
被引数:
18.1W
机构
暂无机构信息
引用论文
500 GHz plasmonic Mach-Zehnder modulator enabling sub-THz microwave photonics实现亚太赫兹微波光子学的500 GHz等离子体马赫-曾德尔调制器.png)
APL PHOTONICSOAAI
APL PHOTONICSIF5.3
OAAI分享收藏Integrating heterogeneous drug sensitivity data from cancer pharmacogenomic studiesOncotargetOAAI
OncotargetIF0
OAAI分享收藏Photon-trapping microstructures enable high-speed high-efficiency silicon photodiodesNATURE PHOTONICS
NATURE PHOTONICSIF32.9
PREAI
分享收藏Freely scalable and reconfigurable optical hardware for deep learning用于深度学习的可自由扩展和可重新配置的光学硬件SCIENTIFIC REPORTSOAAI
SCIENTIFIC REPORTSIF3.9
OAAI分享收藏
查看更多内容