Broadband sensory networks with locally stored responsivities for neuromorphic machine vision

Abstract

As the most promising candidates for the implementation of in-sensor computing, retinomorphic vision sensors can constitute built-in neural networks and directly implement multiply-and-accumulation operations using responsivities as the weights. However, existing retinomorphic vision sensors mainly use a sustained gate bias to maintain the responsivity due to its volatile nature. Here, we propose an ion-induced localized-field strategy to develop retinomorphic vision sensors with nonvolatile tunable responsivity in both positive and negative regimes and construct a broadband and reconfigurable sensory network with locally stored weights to implement in-sensor convolutional processing in spectral range of 400 to 1800 nanometers. In addition to in-sensor computing, this retinomorphic device can implement in-memory computing benefiting from the nonvolatile tunable conductance, and a complete neuromorphic visual system involving front-end in-sensor computing and back-end in-memory computing architectures has been constructed, executing supervised and unsupervised learning tasks as demonstrations. This work paves the way for the development of high-speed and low-power neuromorphic machine vision for time-critical and data-intensive applications.

ZnO Nanoparticle-Induced Performance Enhancement of a Coumarin-Based Nonvolatile Memory Device

“…Hence, in order to assess the device-to-device variability of the switching parameters as well as the reliability of the above results, a total of 36 independent and almost identical memory units (6 × 6 array) like that of D 1 were fabricated. It was found that out of these 36 memory units (cells), only 13 showed reliable WORM behavior with a significant memory window (∼10). Figure e shows the I – V response of these 13 memory units.…”

“…1−3 Conventional memory devices, which are mainly based on silicon, have several limitations such as downscaling, high power consumption, heat death, etc. 2 Moreover, performance enhancement of these devices has become excessively difficult in recent times. 4 Resistive switching memory devices based on various novel and easy-to-process materials can potentially serve as a solution to these problems.…”

“…Over the past few years, resistive switching memory has gained immense popularity within the scientific community owing to its excellent nonvolatility, high endurance, long retention time, fast switching speed, low power consumption, excellent down scalability, and high-density data storage capability, providing a feasible and cost-effective alternative to the existing silicon-based memory devices. − Conventional memory devices, which are mainly based on silicon, have several limitations such as downscaling, high power consumption, heat death, etc . Moreover, performance enhancement of these devices has become excessively difficult in recent times .…”

In-sensor neuromorphic computing using perovskites and transition metal dichalcogenides

“…Besides, up to now, the computing results are generally obtained through simulation based on software. Therefore, achieving neuromorphic computing through extensive integration of arrays in hardware is a challenging journey that lies ahead [282].…”

A Reconfigurable All-Optical-Controlled Synaptic Device for Neuromorphic Computing Applications

“…Nevertheless, due to the fact that optical artificial synapses exhibit a unidirectional response to light stimuli, both optical and electrical coconfigurations are necessary to accomplish reversible modulation of synaptic weights in most of the reported photoelectric synapses. ,,− Moreover, optical stimulation often relates to the potentiation process, whereas electrical stimulation acts on the depression process. Alternatively, an optimal all-optical modulation of artificial synapses might eliminate the need for mixed optoelectronic signals, simplify the configuration of external stimuli, as well as greatly enhance the utilization of optical signals. − The discussed all-optical controlled synaptic devices exhibit an increase and decrease in postsynaptic currents (PSCs) under different light stimuli, corresponding to positive photoconductance (PPC) and negative photoconductance (NPC), respectively. It is obvious that realizing a negative photoresponse is the primary obstacle to exhibiting a switchable bidirectional photoresponse in optoelectronic synapses.…”