Reservoir computing and task performing through using high-$\beta$ lasers with delayed optical feedback
Abstract
Nonlinear photonic sources including semiconductor lasers have recently been utilized as ideal computation elements for information processing. They supply energy-efficient way and rich dynamics for classification and recognition tasks. In this work, we propose and numerically study the dynamics of complex photonic systems including high- laser element with delayed feedback and functional current modulation, and employ nonlinear laser dynamics of near-threshold region for the application in time-delayed reservoir computing. The results indicate a perfect (100) recognition accuracy for the pattern recognition task, and an accuracy of about 98 for the Mackey-Glass chaotic sequences prediction. Therefore, the system shows an improvement of performance with low-power consumption, in particular, the error rate is an order of magnitude smaller in comparison with previous works. Furthermore, by changing the DC pump, we are able to modify the amount of spontaneous emission photons of the system, this then allow us to explore how the laser noise impact the performance of the reservoir computing system. Through manipulating these variables, we show a deeper understanding on the proposed system, which is helpful for the practical applications of reservoir computing.
Keywords
Cite
@article{arxiv.2305.11878,
title = {Reservoir computing and task performing through using high-$\beta$ lasers with delayed optical feedback},
author = {T. Wang and C. Jiang and Q. Fang and X. Guo and Y. Zhang and C. Jin and S. Xiang},
journal= {arXiv preprint arXiv:2305.11878},
year = {2023}
}