TY - GEN
T1 - An Optical Computing Chip for Executing Complex-valued Neural Network and Its On-chip Training
AU - Zhang, Hui
AU - Liu, Ai Qun
N1 - Funding Information:
This work was supported by the Singapore Ministry of Education (MOE) Tier 3 grant (MOE2017-T3-1-001), the Singapore National Research Foundation (NRF) National Natural Science Foundation of China (NSFC) joint grant (NRF2017NRF-NSFC002-014).
Publisher Copyright:
© 2022 IEEE.
PY - 2022/4
Y1 - 2022/4
N2 - The optical implementation of neural networks is proposed to have advantages over electronic implementations with lower power consumption and higher computation speed. However, most optical neural networks (ONNs) utilize conventional real-valued frameworks that are designed for digital computers, forfeiting many advantages of optical computing such as efficient complex-valued operations. Complex-valued neural networks are advantageous to their real-valued counterparts by offering rich representation space, fast convergence, and strong generalizations. We propose and demonstrate an ONN that implements truly complex-valued neural networks, achieving high accuracy and strong learning capability in many benchmark tasks [1]. On the other hand, efficiently training ONNs remains a formidable challenge, due to the difficulty in obtaining gradient information from a physical device. We propose an efficient on-chip training protocol for ONNs and demonstrate it by several practical tasks [2]. The protocol is gradient-free and physical agnostic, and is applicable for various types of chip structures, especially those that cannot be analytically decomposed and characterized. The protocol is robust to experimental perturbations like imperfect phase detection and photodetection noise. Our results present a promising avenue towards deep complex networks with smaller chip size, stronger performance, and flexible reconfiguration to realistic applications (e.g., facial recognition, natural language processing, and autonomous vehicles).
AB - The optical implementation of neural networks is proposed to have advantages over electronic implementations with lower power consumption and higher computation speed. However, most optical neural networks (ONNs) utilize conventional real-valued frameworks that are designed for digital computers, forfeiting many advantages of optical computing such as efficient complex-valued operations. Complex-valued neural networks are advantageous to their real-valued counterparts by offering rich representation space, fast convergence, and strong generalizations. We propose and demonstrate an ONN that implements truly complex-valued neural networks, achieving high accuracy and strong learning capability in many benchmark tasks [1]. On the other hand, efficiently training ONNs remains a formidable challenge, due to the difficulty in obtaining gradient information from a physical device. We propose an efficient on-chip training protocol for ONNs and demonstrate it by several practical tasks [2]. The protocol is gradient-free and physical agnostic, and is applicable for various types of chip structures, especially those that cannot be analytically decomposed and characterized. The protocol is robust to experimental perturbations like imperfect phase detection and photodetection noise. Our results present a promising avenue towards deep complex networks with smaller chip size, stronger performance, and flexible reconfiguration to realistic applications (e.g., facial recognition, natural language processing, and autonomous vehicles).
UR - http://www.scopus.com/inward/record.url?scp=85132772402&partnerID=8YFLogxK
U2 - 10.1109/PIERS55526.2022.9793216
DO - 10.1109/PIERS55526.2022.9793216
M3 - Conference article published in proceeding or book
AN - SCOPUS:85132772402
T3 - Progress in Electromagnetics Research Symposium
SP - 189
EP - 196
BT - 2022 Photonics and Electromagnetics Research Symposium, PIERS 2022 - Proceedings
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 2022 Photonics and Electromagnetics Research Symposium, PIERS 2022
Y2 - 25 April 2022 through 29 April 2022
ER -