Abstract
We propose and experimentally demonstrate a simple and energy-efficient photonic convolutional accelerator based on a monolithically integrated multi-wavelength distributed feedback semiconductor laser using the superimposed sampled Bragg grating structure. The photonic convolutional accelerator operates at 44.48 GOPS for one 2 × 2 kernel with a convolutional window vertical sliding stride of 2 and generates 100 images of real-time recognition. Furthermore, a real-time recognition task on the MNIST database of handwritten digits with a prediction accuracy of 84% is achieved. This work provides a compact and low-cost way to realize photonic convolutional neural networks.
© 2022 Optica Publishing Group
Full Article | PDF ArticleMore Like This
Yuxin Ma, Kaifei Tang, Jun Lu, Zhenxing Sun, Lu Yao, Yujiang Wang, Zhenzhen Xu, Mi Li, Xin Wang, Rulei Xiao, Yu Xin, Jiahui Liu, Xiang Ji, Wei Jiang, Yuechun Shi, and Xiangfei Chen
Opt. Lett. 49(10) 2629-2632 (2024)
Xiangyan Meng, Nuannuan Shi, Difei Shi, Wei Li, and Ming Li
Opt. Express 30(10) 16217-16228 (2022)
Wai Yuen Fu, Yuk Fai Cheung, and Hoi Wai Choi
Opt. Lett. 47(24) 6397-6400 (2022)