Silicon Ring Resonator with Phase-Change Material as a Plastic Dynamical Node for Scalable All-Optical Neural Networks with Synaptic Plasticity

Abstract

Synaptic plasticity, that is the ability of connections in neural networks to strengthen or weaken depending on their input, is a fundamental component of learning and memory in biological brains. We present a numerical and experimental investigation of an integrated photonic plastic node, consisting of a silicon ring resonator enhanced by phase-change materials (GST). This all-optical device is capable of dynamical nonlinear behaviour, multi-scale volatile memory, non-volatile memory and multi-wavelength operations. We propose its employment as a building block in scalable all-optical dynamical neural networks that can adapt to their input via synaptic plasticity.