A Review of Artificial Spiking Neuron Devices for Neural Processing and Sensing

Abstract

A spiking neural network (SNN) inspired by the structure and principles of the human brain can significantly enhance the energy efficiency of artificial intelligence computing by overcoming the bottlenecks of the conventional von Neumann architecture with its massive parallelism and spike transmissions. The construction of artificial neurons is important for the hardware implementation of an SNN, which generates spike signals when enough synaptic signals are gathered. Because circuit‐level artificial neurons with comparator and reset circuits require considerable hardware area, intensive efforts are devoted in recent years for building artificial neurons at the device level for better area efficiency. Furthermore, artificial sensory neuron devices, which perform neural processing and sensing concurrently, have recently been developed in order to reduce the hardware cost and energy consumption of traditional sensory systems through in‐sensor computing. This review article surveys and benchmarks the recent progress of artificial neuron devices for neural processing and sensing. First, various artificial neuron devices are summarized, including single‐transistor neurons (1T‐neurons), memristor neurons, phase‐change neurons, magnetic neurons, and ferroelectric neurons. Next, cointegration technologies with artificial synaptic devices and artificial sensory neurons for in‐sensor computing are introduced. Finally, the challenges and prospects for developing artificial neuron devices are discussed.

Emerging MXene‐Based Memristors for In‐Memory, Neuromorphic Computing, and Logic Operation

“…In this regard, memristors have emerged as promising contenders for artificial synaptic application. [177][178][179] For memristors, the change of conductance during operation can be utilized to emulate the change of synaptic weight (ΔW), which is defined as ΔW = (G post − G pre )/G pre × 100%, where G pre and G post respectively represent the conductance of the memristor in response to the presynaptic and postsynaptic pulses. When Δt (i.e., t post − t pre ) and ΔW are both positive, the electrical potentiation of memristor occurs.…”

Nanomaterial-Based Synaptic Optoelectronic Devices for In-Sensor Preprocessing of Image Data

“…For example, the postsynaptic potential is accumulated by repetitive presynaptic potentials, and the neurons fire the action potential once the postsynaptic potential exceeds the threshold. Such properties can be realized by using external electronic circuits (e.g., comparators and reset circuits) or by integrating two or more optoelectronic components; 131 however, they are not ideal in terms of power consumption and hardware complexity. 132 In this regard, the realization of such properties at the single-device level would be beneficial for energy-efficient machine vision as well as module miniaturization.…”

An Oxide Based Spiking Thermoreceptor for Low-Power Thermography Edge Detection

“…Inspired by the biological sensory processing paradigm, the artificial spiking receptor (ASR) has aroused extensive interesting recently [2], [3], [5], [6], [7], [8], [9], [10], [11]. For example, a spiking photoreceptor based on indium gallium zinc oxide (IGZO) has been proposed with the capability to convert UV light into a spike train at a rate dependent on the light wavelength.…”