Summary
International Conference on Emerging Technologies for Communications
2020
Session Number:I1
Session:
Number:I1-4
Activation function of artificial neural networks for optical nonlinearity compensation
Yuki Miyashita, Takeru Kyono, Kai Ikuta, Yuichiro Kurokawa, Moriya Nakamura,
pp.-
Publication Date:2020/12/2
Online ISSN:2188-5079
DOI:10.34385/proc.63.I1-4
PDF download
Summary:
We investigated the performance of an artificial neural network-based nonlinear equalizer for optical nonlinearity compensation, by comparing a sigmoid function and ReLU as the activation functions. The investigation was performed by numerical simulation of optical fiber transmission of 10-GSymbol/s 16QAM signals. The results show that the sigmoid function had better performance in terms of learning speed.