Analog Nanoscale Electro-Optical Synapses for Neuromorphic Computing Applications

Abstract

The typically nonlinear and asymmetric response of synaptic memristors to positive and negative electrical pulses makes the realization of accurate deep neural networks very challenging. Here, we integrate a two-terminal valence change memory (VCM) into a photonic/plasmonic circuit and show that the switching properties of this memristor become more gradual and symmetric under light irradiation. The added optical input acts on the VCM as a third, independent modulation channel. It locally heats the active area of the device, which enhances the generation of oxygen vacancies and broadens the resulting nanoscale conductive filaments. The measured conductance modulation of the VCM is then inserted into a neural network simulator. Using the MNIST data set of handwritten digits as an application, a light-enhanced recognition accuracy of 93.53% is demonstrated, similar to ideally performing memristors (94.86%) and much higher than those without light (67.37%). Notably, the optical signal does not increase the overall energy consumption by more than 3.2%. Finally, an approach to scale up our electro-optical technology is proposed, which could allow high-density, energy-efficient neuromorphic computing chips.

Biopolymer based artificial synapses enable linear conductance tuning and low-power for neuromorphic computing

“…More programmable states mean the conductance (synaptic weight) within the same dynamic range can be modulated more precisely. 33 Meanwhile a high precision of electronics is extensively demanded in applications such as automatic driving, breakdown diagnosis, and high-resolution displays. 34 The impact of pulse parameters (amplitude, width, and interval) on the conductance of the device was investigated.…”

“…It is noteworthy that an external heat or light source is required to motivate the migration of oxygen vacancies in the previous study. 11,40 However, in our memristor the evenly distributed Ag clusters could be heat sources on their own to promote a stable ion migration process. Herein, a part of the polymer layer containing embedded Ag clusters is conductive and the unembedded part remains insulative (serving as a blocking layer).…”

A Van Der Waals Photo‐Ferroelectric Synapse

“…Thus, by reading I sc , one can extract the associated memory. [ 29 ] In our photo‐ferroelectric synapse, we utilize the photo‐ferroelectric effect for synaptic memory extraction. The key requirement of such synaptic device is that the ferroelectric channel has to be an optically active semiconductor.…”

An Ultra-Compact Design of Plasmonic Memristor with Low Loss and High Extinction Efficiency Based on Enhanced Interaction between Filament and Concentrated Plasmon

“…This system (namely, "plasmonic memristor"), on one hand, has volatile or non-volatile memory with electrical write/erase and optical readout functionality [10,15,16]. On the other artificial synapses, image recognition and even neuromorphic visual systems, were explored at different wavelength ranges [10,14,15,[19][20][21][22][23]. The plasmonic memristor based on a silicon photonics platform is especially attractive, due to its emerging chances to realize large-scale interrogation via a CMOS compatible process [10,14,15,19,20], as well as its potential to fuse its data processing capability with the strong power of silicon photonics on data transmission at communication wavelengths around 1.31 μm and 1.55 μm [19,20].…”