Single perceptron at 12 GigaOPs based on a microcomb for versatile, high-speed scalable, optical neural networks

M. Tan; X. Xu; J. Wu; A. Boes; B. Corcoran; T. G. Nguyen; S. T. Chu; B. E. Little; R. Morandotti; A. Mitchell; D. G. Hicks; D. J. Moss

doi:10.1117/12.2584011

Single perceptron at 12 GigaOPs based on a microcomb for versatile, high-speed scalable, optical neural networks

M. Tan
, X. Xu
, J. Wu
, A. Boes
, B. Corcoran
, T. G. Nguyen
, S. T. Chu
, B. E. Little
, R. Morandotti
, A. Mitchell
, D. G. Hicks
, D. J. Moss

Research output: Chapters, Conference Papers, Creative and Literary Works › RGC 32 - Refereed conference paper (with host publication) › peer-review

Abstract

Optical artificial neural networks (ONNs) have significant potential for ultra-high computing speed and energy efficiency. We report a novel approach to ONNs that uses integrated Kerr optical micro-combs. This approach is programmable and scalable and is capable of reaching ultra-high speeds. We demonstrate the basic building block ONNs- A single neuron perceptron-by mapping synapses onto 49 wavelengths to achieve an operating speed of 11.9 x 109 operations per second, or Giga-OPS, at 8 bits per operation, which equates to 95.2 gigabits/s (Gbps). We test the perceptron on handwritten-digit recognition and cancer-cell detection- A chieving over 90% and 85% accuracy, respectively. By scaling the perceptron to a deep learning network using off-the-shelf telecom technology we can achieve high throughput operation for matrix multiplication for real-time massive data processing.

Original language	English
Title of host publication	Smart Photonic and Optoelectronic Integrated Circuits XXIII
Editors	Sailing He, Laurent Vivien
Publisher	SPIE
ISBN (Electronic)	9781510642164
ISBN (Print)	9781510642157
DOIs	https://doi.org/10.1117/12.2584011
Publication status	Published - Mar 2021
Event	Smart Photonic and Optoelectronic Integrated Circuits XXIII - Virtual, United States Duration: 6 Mar 2021 → 11 Mar 2021

Publication series

Name	Proceedings of SPIE - The International Society for Optical Engineering
Volume	11690
ISSN (Print)	0277-786X
ISSN (Electronic)	1996-756X

Conference

Conference	Smart Photonic and Optoelectronic Integrated Circuits XXIII
Place	United States
Period	6/03/21 → 11/03/21

Bibliographical note

Full text of this publication does not contain sufficient affiliation information. With consent from the author(s) concerned, the Research Unit(s) information for this record is based on the existing academic department affiliation of the author(s).

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

SDG 3 Good Health and Well-being
SDG 7 Affordable and Clean Energy

Research Keywords

Kerr micro-combs
micro-ring resonators
microwave photonics
Optical neural networks

Access Output

10.1117/12.2584011

Other Access Options

Check CityUHK Library

Permanent Link

Right click to copy link

Cite this

Tan, M., Xu, X., Wu, J., Boes, A., Corcoran, B., Nguyen, T. G., Chu, S. T., Little, B. E., Morandotti, R., Mitchell, A., Hicks, D. G., & Moss, D. J. (2021). Single perceptron at 12 GigaOPs based on a microcomb for versatile, high-speed scalable, optical neural networks. In S. He, & L. Vivien (Eds.), Smart Photonic and Optoelectronic Integrated Circuits XXIII Article 116900M (Proceedings of SPIE - The International Society for Optical Engineering; Vol. 11690). SPIE. https://doi.org/10.1117/12.2584011

@inproceedings{7c08697d8c7e41abbe99a84a30699217,

title = "Single perceptron at 12 GigaOPs based on a microcomb for versatile, high-speed scalable, optical neural networks",

abstract = "Optical artificial neural networks (ONNs) have significant potential for ultra-high computing speed and energy efficiency. We report a novel approach to ONNs that uses integrated Kerr optical micro-combs. This approach is programmable and scalable and is capable of reaching ultra-high speeds. We demonstrate the basic building block ONNs- A single neuron perceptron-by mapping synapses onto 49 wavelengths to achieve an operating speed of 11.9 x 109 operations per second, or Giga-OPS, at 8 bits per operation, which equates to 95.2 gigabits/s (Gbps). We test the perceptron on handwritten-digit recognition and cancer-cell detection- A chieving over 90\% and 85\% accuracy, respectively. By scaling the perceptron to a deep learning network using off-the-shelf telecom technology we can achieve high throughput operation for matrix multiplication for real-time massive data processing.",

keywords = "Kerr micro-combs, micro-ring resonators, microwave photonics, Optical neural networks",

author = "M. Tan and X. Xu and J. Wu and A. Boes and B. Corcoran and Nguyen, \{T. G.\} and Chu, \{S. T.\} and Little, \{B. E.\} and R. Morandotti and A. Mitchell and Hicks, \{D. G.\} and Moss, \{D. J.\}",

note = "Full text of this publication does not contain sufficient affiliation information. With consent from the author(s) concerned, the Research Unit(s) information for this record is based on the existing academic department affiliation of the author(s).; Smart Photonic and Optoelectronic Integrated Circuits XXIII ; Conference date: 06-03-2021 Through 11-03-2021",

year = "2021",

month = mar,

doi = "10.1117/12.2584011",

language = "English",

isbn = "9781510642157",

series = "Proceedings of SPIE - The International Society for Optical Engineering",

publisher = "SPIE",

editor = "Sailing He and Laurent Vivien",

booktitle = "Smart Photonic and Optoelectronic Integrated Circuits XXIII",

address = "United States",

}

Tan, M, Xu, X, Wu, J, Boes, A, Corcoran, B, Nguyen, TG, Chu, ST, Little, BE, Morandotti, R, Mitchell, A, Hicks, DG & Moss, DJ 2021, Single perceptron at 12 GigaOPs based on a microcomb for versatile, high-speed scalable, optical neural networks. in S He & L Vivien (eds), Smart Photonic and Optoelectronic Integrated Circuits XXIII., 116900M, Proceedings of SPIE - The International Society for Optical Engineering, vol. 11690, SPIE, Smart Photonic and Optoelectronic Integrated Circuits XXIII, United States, 6/03/21. https://doi.org/10.1117/12.2584011

Single perceptron at 12 GigaOPs based on a microcomb for versatile, high-speed scalable, optical neural networks. / Tan, M.; Xu, X.; Wu, J. et al.
Smart Photonic and Optoelectronic Integrated Circuits XXIII. ed. / Sailing He; Laurent Vivien. SPIE, 2021. 116900M (Proceedings of SPIE - The International Society for Optical Engineering; Vol. 11690).

Research output: Chapters, Conference Papers, Creative and Literary Works › RGC 32 - Refereed conference paper (with host publication) › peer-review

TY - GEN

T1 - Single perceptron at 12 GigaOPs based on a microcomb for versatile, high-speed scalable, optical neural networks

AU - Tan, M.

AU - Xu, X.

AU - Wu, J.

AU - Boes, A.

AU - Corcoran, B.

AU - Nguyen, T. G.

AU - Chu, S. T.

AU - Little, B. E.

AU - Morandotti, R.

AU - Mitchell, A.

AU - Hicks, D. G.

AU - Moss, D. J.

N1 - Full text of this publication does not contain sufficient affiliation information. With consent from the author(s) concerned, the Research Unit(s) information for this record is based on the existing academic department affiliation of the author(s).

PY - 2021/3

Y1 - 2021/3

N2 - Optical artificial neural networks (ONNs) have significant potential for ultra-high computing speed and energy efficiency. We report a novel approach to ONNs that uses integrated Kerr optical micro-combs. This approach is programmable and scalable and is capable of reaching ultra-high speeds. We demonstrate the basic building block ONNs- A single neuron perceptron-by mapping synapses onto 49 wavelengths to achieve an operating speed of 11.9 x 109 operations per second, or Giga-OPS, at 8 bits per operation, which equates to 95.2 gigabits/s (Gbps). We test the perceptron on handwritten-digit recognition and cancer-cell detection- A chieving over 90% and 85% accuracy, respectively. By scaling the perceptron to a deep learning network using off-the-shelf telecom technology we can achieve high throughput operation for matrix multiplication for real-time massive data processing.

AB - Optical artificial neural networks (ONNs) have significant potential for ultra-high computing speed and energy efficiency. We report a novel approach to ONNs that uses integrated Kerr optical micro-combs. This approach is programmable and scalable and is capable of reaching ultra-high speeds. We demonstrate the basic building block ONNs- A single neuron perceptron-by mapping synapses onto 49 wavelengths to achieve an operating speed of 11.9 x 109 operations per second, or Giga-OPS, at 8 bits per operation, which equates to 95.2 gigabits/s (Gbps). We test the perceptron on handwritten-digit recognition and cancer-cell detection- A chieving over 90% and 85% accuracy, respectively. By scaling the perceptron to a deep learning network using off-the-shelf telecom technology we can achieve high throughput operation for matrix multiplication for real-time massive data processing.

KW - Kerr micro-combs

KW - micro-ring resonators

KW - microwave photonics

KW - Optical neural networks

UR - https://www.scopus.com/pages/publications/85105938940

UR - https://www.scopus.com/record/pubmetrics.uri?eid=2-s2.0-85105938940&origin=recordpage

U2 - 10.1117/12.2584011

DO - 10.1117/12.2584011

M3 - RGC 32 - Refereed conference paper (with host publication)

SN - 9781510642157

T3 - Proceedings of SPIE - The International Society for Optical Engineering

BT - Smart Photonic and Optoelectronic Integrated Circuits XXIII

A2 - He, Sailing

A2 - Vivien, Laurent

PB - SPIE

T2 - Smart Photonic and Optoelectronic Integrated Circuits XXIII

Y2 - 6 March 2021 through 11 March 2021

ER -

Tan M, Xu X, Wu J, Boes A, Corcoran B, Nguyen TG et al. Single perceptron at 12 GigaOPs based on a microcomb for versatile, high-speed scalable, optical neural networks. In He S, Vivien L, editors, Smart Photonic and Optoelectronic Integrated Circuits XXIII. SPIE. 2021. 116900M. (Proceedings of SPIE - The International Society for Optical Engineering). doi: 10.1117/12.2584011

Single perceptron at 12 GigaOPs based on a microcomb for versatile, high-speed scalable, optical neural networks

Abstract

Publication series

Conference

Bibliographical note

UN SDGs

Research Keywords

Access Output

Other Access Options

Permanent Link

Other Files and Links

Fingerprint

Cite this