Poster + Paper
9 October 2021 All-optical nonlinear activation functions based on germanium silicon hybrid integration
Author Affiliations +
Conference Poster
Abstract
Optical neural networks (ONNs), a kind of neural network implemented by optical hardware, have attracted more and more attention due to their excellent performance. As a part of ONNs, nonlinear activation functions the key for ONN to achieve more various functions. The realization of nonlinear activation functions by optical hardware has become a research hotspot. Here, we propose and demonstrate an all-optical implementation of a nonlinear activation function, which utilizes the intrinsic absorption and the plasma dispersion effect of germanium at 1550nm. The one has a loss about 4db and a threshold power of 5.1mW, while the another has a loss around 11db and a threshold power of 56.2mW. This proves that our scheme has great potential for nonlinear activation functions.
© (2021) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Hengkang Li and Jianji Dong "All-optical nonlinear activation functions based on germanium silicon hybrid integration", Proc. SPIE 11894, Optoelectronic Devices and Integration X, 1189413 (9 October 2021); https://doi.org/10.1117/12.2602448
Advertisement
Advertisement
RIGHTS & PERMISSIONS
Get copyright permission  Get copyright permission on Copyright Marketplace
KEYWORDS
Germanium

Waveguides

Silicon

Plasma

Absorption

Refractive index

Electron beam lithography

RELATED CONTENT


Back to Top