The original paper is in English. Non-English content has been machine-translated and may contain typographical errors or mistranslations. ex. Some numerals are expressed as "XNUMX".
Copyrights notice
The original paper is in English. Non-English content has been machine-translated and may contain typographical errors or mistranslations. Copyrights notice
집적 나노포토닉스 기술의 급속한 발전으로 인해 광신경망 구조가 폭넓게 연구되고 있다. 광신경망은 광신호를 네트워크에 전파하는 것만으로도 추론 처리를 완료할 수 있기 때문에 전자공학만으로 구현한 인공신경망(ANN)보다 XNUMX배 이상 빠른 속도가 기대된다. 본 논문에서는 초광대역으로 빛의 속도로 추론 처리를 가능하게 하는 파장 분할 다중화를 이용한 광 벡터 행렬 곱셈(VMM) 회로를 먼저 제안한다. 다음으로 본 논문에서는 속도 성능을 희생하지 않고 추론 처리의 정확성을 크게 향상시키는 일괄 정규화 및 활성화 기능을 위한 광전자 회로 구현을 제안합니다. 마지막으로 기계 학습을 위한 가상 환경과 광전자 회로 시뮬레이터를 사용하여 광전자 ANN 회로의 초고속 및 정확한 작동을 시연합니다.
Naoki HATTORI
Nagoya University
Jun SHIOMI
Kyoto University
Yutaka MASUDA
Nagoya University
Tohru ISHIHARA
Nagoya University
Akihiko SHINYA
NTT Nanophotonics Center,NTT Basic Research Laboratories
Masaya NOTOMI
NTT Nanophotonics Center,NTT Basic Research Laboratories
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
부
Naoki HATTORI, Jun SHIOMI, Yutaka MASUDA, Tohru ISHIHARA, Akihiko SHINYA, Masaya NOTOMI, "Neural Network Calculations at the Speed of Light Using Optical Vector-Matrix Multiplication and Optoelectronic Activation" in IEICE TRANSACTIONS on Fundamentals,
vol. E104-A, no. 11, pp. 1477-1487, November 2021, doi: 10.1587/transfun.2020KEP0016.
Abstract: With the rapid progress of the integrated nanophotonics technology, the optical neural network architecture has been widely investigated. Since the optical neural network can complete the inference processing just by propagating the optical signal in the network, it is expected more than one order of magnitude faster than the electronics-only implementation of artificial neural networks (ANN). In this paper, we first propose an optical vector-matrix multiplication (VMM) circuit using wavelength division multiplexing, which enables inference processing at the speed of light with ultra-wideband. This paper next proposes optoelectronic circuit implementation for batch normalization and activation function, which significantly improves the accuracy of the inference processing without sacrificing the speed performance. Finally, using a virtual environment for machine learning and an optoelectronic circuit simulator, we demonstrate the ultra-fast and accurate operation of the optical-electronic ANN circuit.
URL: https://global.ieice.org/en_transactions/fundamentals/10.1587/transfun.2020KEP0016/_p
부
@ARTICLE{e104-a_11_1477,
author={Naoki HATTORI, Jun SHIOMI, Yutaka MASUDA, Tohru ISHIHARA, Akihiko SHINYA, Masaya NOTOMI, },
journal={IEICE TRANSACTIONS on Fundamentals},
title={Neural Network Calculations at the Speed of Light Using Optical Vector-Matrix Multiplication and Optoelectronic Activation},
year={2021},
volume={E104-A},
number={11},
pages={1477-1487},
abstract={With the rapid progress of the integrated nanophotonics technology, the optical neural network architecture has been widely investigated. Since the optical neural network can complete the inference processing just by propagating the optical signal in the network, it is expected more than one order of magnitude faster than the electronics-only implementation of artificial neural networks (ANN). In this paper, we first propose an optical vector-matrix multiplication (VMM) circuit using wavelength division multiplexing, which enables inference processing at the speed of light with ultra-wideband. This paper next proposes optoelectronic circuit implementation for batch normalization and activation function, which significantly improves the accuracy of the inference processing without sacrificing the speed performance. Finally, using a virtual environment for machine learning and an optoelectronic circuit simulator, we demonstrate the ultra-fast and accurate operation of the optical-electronic ANN circuit.},
keywords={},
doi={10.1587/transfun.2020KEP0016},
ISSN={1745-1337},
month={November},}
부
TY - JOUR
TI - Neural Network Calculations at the Speed of Light Using Optical Vector-Matrix Multiplication and Optoelectronic Activation
T2 - IEICE TRANSACTIONS on Fundamentals
SP - 1477
EP - 1487
AU - Naoki HATTORI
AU - Jun SHIOMI
AU - Yutaka MASUDA
AU - Tohru ISHIHARA
AU - Akihiko SHINYA
AU - Masaya NOTOMI
PY - 2021
DO - 10.1587/transfun.2020KEP0016
JO - IEICE TRANSACTIONS on Fundamentals
SN - 1745-1337
VL - E104-A
IS - 11
JA - IEICE TRANSACTIONS on Fundamentals
Y1 - November 2021
AB - With the rapid progress of the integrated nanophotonics technology, the optical neural network architecture has been widely investigated. Since the optical neural network can complete the inference processing just by propagating the optical signal in the network, it is expected more than one order of magnitude faster than the electronics-only implementation of artificial neural networks (ANN). In this paper, we first propose an optical vector-matrix multiplication (VMM) circuit using wavelength division multiplexing, which enables inference processing at the speed of light with ultra-wideband. This paper next proposes optoelectronic circuit implementation for batch normalization and activation function, which significantly improves the accuracy of the inference processing without sacrificing the speed performance. Finally, using a virtual environment for machine learning and an optoelectronic circuit simulator, we demonstrate the ultra-fast and accurate operation of the optical-electronic ANN circuit.
ER -