Abstract

We experimentally demonstrate two types of programmable, low-threshold, optically controlled nonlinear activation functions, which are challenging to realize in photonic neural networks (PNNs). These devices rely on on-chip integrated Ge–Si photoelectric detectors and silicon electro-optical switches, and they generate rectified linear unit (ReLU) or sigmoid functions with arbitrary slopes without additional electrical processing. Both devices function at an extremely low threshold of 0.2 mW. The embedding of these nonlinear activation functions into convolutional neural networks facilitates the attainment of high inference accuracies of up to 95% when applied to Modified National Institute of Standards and Technology (MNIST) handwritten digit-classification tasks. The devices are suitable for low-power PNNs with an arbitrary number of propagation layers in photonic-computing chips.

© 2022 Optica Publishing Group

Reconfigurable all-optical nonlinear activation functions for neuromorphic photonics

Aashu Jha, Chaoran Huang, and Paul R. Prucnal
Opt. Lett. 45(17) 4819-4822 (2020)

Optical ReLU using membrane lasers for an all-optical neural network

Naoki Takahashi, Weicheng Fang, Ruihao Xue, Sho Okada, Yoshitaka Ohiso, Tomohiro Amemiya, and Nobuhiko Nishiyama
Opt. Lett. 47(21) 5715-5718 (2022)

Low-threshold all-optical nonlinear activation function based on a Ge/Si hybrid structure in a microring resonator

Bo Wu, Hengkang Li, Weiyu Tong, Jianji Dong, and Xinliang Zhang
Opt. Mater. Express 12(3) 970-980 (2022)

Programmable low-power consumption all-optical nonlinear activation functions using a micro-ring resonator with phase-change materials

Ziling Fu, Zhi Wang, Peter Bienstman, Rui Jiang, Jian Wang, and Chongqing Wu
Opt. Express 30(25) 44943-44953 (2022)

Low-threshold all-optical nonlinear activation function based on injection locking in distributed feedback laser diodes

Binshuo Liu, Shaofu Xu, Bowen Ma, Sicheng Yi, and Weiwen Zou
Opt. Lett. 48(15) 3889-3892 (2023)