Photonic multiplexing architectures for optical neuromorphic computation

Abstract

The simultaneous advances in artificial neural networks and photonic integration technologies have spurred extensive research in optical computing and optical neural networks (ONNs). The potential to simultaneously exploit multiple physi-cal dimensions of time, wavelength and space give ONNs the ability to achieve computing operations with high parallel-ism and large-data throughput. Different photonic multiplexing techniques based on these multiple degrees of freedom have enabled ONNs with large-scale interconnectivity and linear computing functions. Here, we review the recent advanc-es of ONNs based on different approaches to photonic multiplexing, and present our outlook on key technologies needed to further advance these photonic multiplexing/hybrid-multiplexing techniques of ONNs.

Review of Integrated Diffractive Deep Neural Networks

“…The two latent variables α 𝑖𝑖 𝑙𝑙 and 𝜙𝜙 𝑖𝑖 𝑙𝑙 should be confined to the range (0, 1) and (0, 2𝜋𝜋) respectively. The forward propagation of the field between two successive layers in 𝐷𝐷 2 𝑁𝑁𝑁𝑁 can be calculated by matrix multiplication [5,24]:…”

Proposal for a monolithic silicon-on-insulator resonator spiking neuron

“…Spiking neural networks (SNNs) in biology encode information through timed events or spikes, utilizing distributed, sparse, and robust coding schemes. Neuromorphic systems are designed to implement artiőcial neural networks (ANNs) with individual neurons in hardware, aiming for enhanced speed, latency, and energy efficiency [1]. Here, we introduce a photonic foundrycompatible, light-based silicon-on-insulator excitable neuron.…”