Optical neural network based on synthetic photonic lattices

Abstract

We reveal that synthetic photonic lattices based on coupled fiber loops can realize deep neural networks for coherent manipulation of optical pulse trains, which could lead to advances in communications and real-Time data analysis. We show the platform feasibility for an essential task of telecommunication signal equalization. We also demonstrated that utilizing all-optical Kerr nonlinearity within the system may open a path to complex signal manipulation. Show more