Digital optical phase and amplitude matrix multiplication processor for neural networks

Xiansong Meng; Kwangwoong Kim; Po Dong; Deming Kong; Hao Hu

doi:10.1364/CLEO_SI.2024.SF2M.4

Digital optical phase and amplitude matrix multiplication processor for neural networks

Xiansong Meng, Kwangwoong Kim, Po Dong, Deming Kong^*, Hao Hu^*

^*Corresponding author for this work

Nokia

Research output: Contribution to conference › Paper › Research › peer-review

Abstract

We propose a high-precision digital optical matrix multiplier utilizing phase and amplitude for neural networks. Results show error-free performance with 16-bit precision in high-definition image processing and no accuracy loss in handwritten digit recognition task.

Original language	English
Publication date	2024
Number of pages	2
DOIs	https://doi.org/10.1364/CLEO_SI.2024.SF2M.4
Publication status	Published - 2024
Event	CLEO: Science and Innovations 2024 - Charlotte Convention Center, Charlotte, United States Duration: 5 May 2024 → 10 May 2024

Conference

Conference	CLEO: Science and Innovations 2024
Location	Charlotte Convention Center
Country/Territory	United States
City	Charlotte
Period	05/05/2024 → 10/05/2024

Keywords

Accuracy
Adaptive optics
Electro-optic effects
Handwriting recognition
Image recognition
Neural networks
Optical computing
Optical fiber networks
Optical imaging
Optical losses

Access to Document

10.1364/CLEO_SI.2024.SF2M.4

OpenUrl availability

Full text

Cite this

@conference{3ca9d2ce32434475962fc47ca2976486,

title = "Digital optical phase and amplitude matrix multiplication processor for neural networks",

abstract = "We propose a high-precision digital optical matrix multiplier utilizing phase and amplitude for neural networks. Results show error-free performance with 16-bit precision in high-definition image processing and no accuracy loss in handwritten digit recognition task.",

keywords = "Accuracy, Adaptive optics, Electro-optic effects, Handwriting recognition, Image recognition, Neural networks, Optical computing, Optical fiber networks, Optical imaging, Optical losses",

author = "Xiansong Meng and Kwangwoong Kim and Po Dong and Deming Kong and Hao Hu",

note = "Publisher Copyright: {\textcopyright} Optica Publishing Group 2024, {\textcopyright} 2024 The Author(s); CLEO: Science and Innovations 2024, CLEO 2024 ; Conference date: 05-05-2024 Through 10-05-2024",

year = "2024",

doi = "10.1364/CLEO_SI.2024.SF2M.4",

language = "English",

}

TY - CONF

T1 - Digital optical phase and amplitude matrix multiplication processor for neural networks

AU - Meng, Xiansong

AU - Kim, Kwangwoong

AU - Dong, Po

AU - Kong, Deming

AU - Hu, Hao

PY - 2024

Y1 - 2024

N2 - We propose a high-precision digital optical matrix multiplier utilizing phase and amplitude for neural networks. Results show error-free performance with 16-bit precision in high-definition image processing and no accuracy loss in handwritten digit recognition task.

AB - We propose a high-precision digital optical matrix multiplier utilizing phase and amplitude for neural networks. Results show error-free performance with 16-bit precision in high-definition image processing and no accuracy loss in handwritten digit recognition task.

KW - Accuracy

KW - Adaptive optics

KW - Electro-optic effects

KW - Handwriting recognition

KW - Image recognition

KW - Neural networks

KW - Optical computing

KW - Optical fiber networks

KW - Optical imaging

KW - Optical losses

U2 - 10.1364/CLEO_SI.2024.SF2M.4

DO - 10.1364/CLEO_SI.2024.SF2M.4

M3 - Paper

AN - SCOPUS:85210502586

T2 - CLEO: Science and Innovations 2024

Y2 - 5 May 2024 through 10 May 2024

ER -

Digital optical phase and amplitude matrix multiplication processor for neural networks

Abstract

Conference

Keywords

Access to Document

OpenUrl availability

Fingerprint

Cite this