Nonlinear synthetic photonic lattices for optical neural networks

Abstract

We reveal that a synthetic photonic lattice based on coupled waveguide loops can serve as a feed-forward neural network for processing of optical pulse sequences in time domain. We demonstrate the capabilities of the system in solving three different problems. We perform efficient training for dispersion compensation for the pulse train distorted in a linear fiber-optic link. We also show that the optical network with an intrinsic Kerr-type nonlinearity can perform a flexible nonlinear transformation of the pulse amplitude as well as discrimination of different pulse sequences. Show more