Photonic In-Memory Computing Primitive for Spiking Neural Networks Using Phase-Change Materials

Abstract

Spiking Neural Networks (SNNs) offer an event-driven and more biologically realistic alternative to standard Artificial Neural Networks based on analog information processing. This can potentially enable energy-efficient hardware implementations of neuromorphic systems which emulate the functional units of the brain, namely, neurons and synapses. Recent demonstrations of ultra-fast photonic computing devices based on phase-change materials (PCMs) show promise of addressing limitations of electrically driven neuromorphic systems. However, scaling these standalone computing devices to a parallel in-memory computing primitive is a challenge. In this work, we utilize the optical properties of the PCM, Ge2Sb2Te5 (GST), to propose a Photonic Spiking Neural Network computing primitive, comprising of a non-volatile synaptic array integrated seamlessly with previously explored 'integrate-and-fire' neurons. The proposed design realizes an 'in-memory' computing platform that leverages the inherent parallelism of wavelength-division-multiplexing (WDM). We show that the proposed computing platform can be used to emulate a SNN inferencing engine for image classification tasks. The proposed design not only bridges the gap between isolated computing devices and parallel large-scale implementation, but also paves the way for ultra-fast computing and localized on-chip learning.

A New Family of Ultralow Loss Reversible Phase‐Change Materials for Photonic Integrated Circuits: Sb2S3 and Sb2Se3

“…However, even the amorphous phase has strong absorption, severely limiting these devices in size and scalability, while the absorption in the crystalline state constricts their use to amplitude modulators rather than to switches and routers. Recent work has indicated how cascading these devices could be used to build a neural network, [ 5,34 ] but the cumulative loss of each device would require frequent amplification in a circuit. Most recently Ge₂Sb₂Se₄Te₁ (GSST) [ 35 ] has been proposed as a low‐loss alternative to GST, where Te is partly substituted by Se.…”

Programmable electronic synapse and nonvolatile resistive switches using MoS2 quantum dots

“…Department of Physics, Indian Institute of Space-Science and Technology (IIST), Valiyamala, Thiruvananthapuram 695547, Kerala, India. * email: kbjinesh@iist.ac.in Interestingly, a large class of nanomaterials has been investigated to understand their neuromorphic responses, because nanostructures have a large number of surface defects that can be electrically modulated [4][5][6][7] .…”

Recent Progress in Synaptic Devices Based on 2D Materials

“…Various synaptic devices based on resistive switching, driven by different physical working mechanisms such as active metallic filament, charge trapping/detrapping effect, ions/vacancies migration, phase change behaviors, ferroelectric polarization, and spin‐transfer torque‐based synapses, have been demonstrated for emerging memory and neuromorphic computing. Many scientists are actively working to resolve various issues in those synaptic devices: high energy consumption, low switching speed, poor reliability, or the lack of high device density for integration.…”