Scalable wavelength-multiplexing photonic reservoir computing

Abstract

Photonic reservoir computing (PRC) is a special hardware recurrent neural network, which is featured with fast training speed and low training cost. This work shows a wavelength-multiplexing PRC architecture, taking advantage of the numerous longitudinal modes in a Fabry–Perot (FP) semiconductor laser. These modes construct connected physical neurons in parallel, while an optical feedback loop provides interactive virtual neurons in series. We experimentally demonstrate a four-channel wavelength-multiplexing PRC architecture with a total of 80 neurons. The clock rate of the multiplexing PRC reaches as high as 1.0 GHz, which is four times higher than that of the single-channel case. In addition, it is proved that the multiplexing PRC exhibits a superior performance on the task of signal equalization in an optical fiber communication link. This improved performance is owing to the rich neuron interconnections both in parallel and in series. In particular, this scheme is highly scalable owing to the rich mode resources in FP lasers.