Spike‐Enabled Audio Learning in Multilevel Synaptic Memristor Array‐Based Spiking Neural Network

Abstract

Speech recognition involves the ability to learn the audios which are closely related to event sequence. Although speech recognition has been widely implemented in software neural networks, a hardware implementation based on energy efficient computing architecture is still missing. Herein, W/MgO/SiO2/Mo memristor arrays with multilevel resistance states are fabricated, where the weights of the artificial synapses in the memristor array can be tuned precisely by voltage pulses. Based on the array, speech recognition in memristive spiking neural networks (SNNs) with improved supervised tempotron algorithm on Texas Instruments digit sequences (TIDIGITS) dataset is conducted, demonstrating software‐comparable accuracy for speech recognition in the memristive SNN. It is envisioned that such memristive SNNs can pave the way to building a bioinspired spike‐based neuromorphic system for audio learning.

Post-silicon nano-electronic device and its application in brain-inspired chips

“…Although these devices can all reach tens or hundreds of resistive states, there are still non-linearities and asymmetries in the tuning. They used W/MgO/SiO 2 /Mo memristive device as the synapse of speech recognition and completed the hardware implementation of SNN using the improved supervised tempotron algorithm on the TIDIGITS dataset (Al-Shedivat et al, 2015 ; Wu et al, 2022 ).…”

Ultrasensitive Flexible Memory Phototransistor with Detectivity of 1.8×1013 Jones for Artificial Visual Nociceptor

“…[20][21][22][23][24][25][26] Since a photon normally only excites one pair of electrons and holes which can be further trapped by the charge-trapping sites, these optoelectronic memory devices are usually based on the continuous photoconductivity effect or light-induced conductivity enhancement. [27][28][29][30] Utilization of heterostructured materials is an effective approach to form a potential barrier to separate photo-induced electrons or holes, accumulate photocarriers at the interface, and subsequently control the channel conductivity, allowing the simulation of ultraweak light-modulated nociceptor by the accumulation of optical signals. [31] Black phosphorous (BP) is a promising candidate for optoelectronic applications owing to its unique anisotropic puckered structure and direct bandgap from 0.3 eV for bulk to 2 eV for monolayer.…”

A bioinspired configurable cochlea based on memristors

“…Although memristor-based Sallen-Key circuit with tunable gain-bandwidth and center frequency characteristics have been proposed for emulating the cochlea, it still lack experimental demonstration based on memristors’ multi-levels ( Li et al, 2020 ; Barraj et al, 2021 ; Onyejegbu et al, 2022 ). In addition, Wu et al (2021) used the stochastic gradient descent-supervised learning rule to train the preprocessed audio features. The weights were mapped into W/MgO/SiO 2 /Mo memristor arrays to complete the speech classification tasks.…”