Thresholding Computing with Heterogeneous Integration of Memristive Kernel with Metal-Oxide-Semiconductor Capacitor for Temporal Data Analysis.

Precise event detection within time-series data is increasingly critical, particularly in noisy environments. Reservoir computing, a robust computing method widely utilized with memristive devices, is efficient in processing temporal signals. However, it typically lacks intrinsic thresholding mechanisms essential for precise event detection. This study introduces a new approach by integrating two Pt/HfO2/TiN (PHT) memristors and one Ni/HfO2/n-Si (NHS) metal-oxide-semiconductor capacitor (2M1MOS) to implement a tunable thresholding function. The current-voltage nonlinearity of memristors combined with the capacitance-voltage nonlinearity of the capacitor forms the basis of the 2M1MOS kernel system. The proposed kernel hardware effectively records feature-specified information of the input signal onto the memristors through capacitive thresholding. In electrocardiogram analysis, the memristive response exhibited a more than ten-fold difference between arrhythmia and normal beats. In isolated spoken digit classification, the kernel achieved an error rate of only 0.7% by tuning thresholds for various time-specific conditions. The kernel is also applied to biometric authentication by extracting personal features using various threshold times, presenting more complex and multifaceted uses of heartbeats and voice data as bio-indicators. These demonstrations highlight the potential of thresholding computing in a memristive framework with heterogeneous integration.

Memory-efficient neurons and synapses for spike-timing-dependent-plasticity in large-scale spiking networks.

This paper addresses the challenges posed by frequent memory access during simulations of large-scale spiking neural networks involving synaptic plasticity. We focus on the memory accesses performed during a common synaptic plasticity rule since this can be a significant factor limiting the efficiency of the simulations. We propose neuron models that are represented by only three state variables, which are engineered to enforce the appropriate neuronal dynamics. Additionally, memory retrieval is executed solely by fetching postsynaptic variables, promoting a contiguous memory storage and leveraging the capabilities of burst mode operations to reduce the overhead associated with each access. Different plasticity rules could be implemented despite the adopted simplifications, each leading to a distinct synaptic weight distribution (i.e., unimodal and bimodal). Moreover, our method requires fewer average memory accesses compared to a naive approach. We argue that the strategy described can speed up memory transactions and reduce latencies while maintaining a small memory footprint.

A Flexible Laser-Induced Graphene Memristor with Volatile Switching for Neuromorphic Applications.

Two-dimensional graphene and graphene-based materials are attracting increasing interest in neuromorphic computing applications by the implementation of memristive architectures that enable the closest solid-state equivalent to biological synapses and neurons. However, the state-of-the-art fabrication methodology involves routine use of high-temperature processes and multistepped chemical synthesis, often on a rigid substrate constraining the experimental exploration in the field to high-tech facilities. Here, we demonstrate the use of a one-step process using a commercial laser to fabricate laser-induced graphene (LIG) memristors directly on a flexible polyimide substrate. For the first time, a volatile resistive switching phenomenon is reported in the LIG without using any additional materials. The absence of any precursor or patterning mask greatly simplifies the process while reducing the cost and providing greater controllability. The fabricated memristors show multilevel resistance-switching characteristics with high endurance and tunable timing characteristics. The recovery time and the trigger pulse-dependent state change are shown to be highly suitable for its use as a synaptic element and in the realization of leaky-integrate and fire neuron in neuromorphic circuits.

Reconfigurable Multilevel Storage and Neuromorphic Computing Based on Multilayer Phase-Change Memory.

In the era of big data, the amount of global data is increasing exponentially, and the storage and processing of massive data put forward higher requirements for memory. To deal with this challenge, high-density memory and neuromorphic computing have been widely investigated. Here, a gradient-doped multilayer phase-change memory with two-level states, four-level states, and linear conductance evolution using different pulse operations is proposed. The mechanism of multilevel states is revealed through high-resolution transmission electron microscopy (HRTEM) and finite-element analysis (FEA), which show that the sequential phase change among different sublayers is realized due to the different physical properties of the sublayers with different doping concentrations. Taking advantage of the devices' linear conductance evolution characteristic, a handwritten digit (28 × 28 pixel) recognition task is implemented with a high learning accuracy of 93.46% by building a simulated artificial neural network made up of this gradient-doped multilayer phase-change memory. It is proved that this gradient-doped multilayer phase-change memory is capable of both binary multilevel digital storage and brain-inspired analog in-memory computing in the same device, enabling reconfigurable applications in the future.

Object Motion Detection Enabled by Reconfigurable Neuromorphic Vision Sensor under Ferroelectric Modulation.

Increasing the demand for object motion detection (OMD) requires shifts of reducing redundancy, heightened power efficiency, and precise programming capabilities to ensure consistency and accuracy. Drawing inspiration from object motion-sensitive ganglion cells, we propose an OMD vision sensor with a simple device structure of a WSe2 homojunction modulated by a ferroelectric copolymer. Under optical mode and intermediate ferroelectric modulation, the vision sensor can generate progressive and bidirectional photocurrents with discrete multistates under zero power consumption. This design enables reconfigurable devices to emulate long-term potentiation and depression for synaptic weights updating, which exhibit 82 states (more than 6 bits) with a uniform step of 6 pA. Such OMD devices also demonstrate nonvolatility, reversibility, symmetry, and ultrahigh linearity, achieving a fitted R2 of 0.999 and nonlinearity values of 0.01/-0.01. Thus, a vision sensor could implement motion detection by sensing only dynamic information based on the brightness difference between frames, while eliminating redundant data from static scenes. Additionally, the neural network utilizing a linear result can recognize the essential moving information with a high recognition accuracy of 96.8%. We also present the scalable potential via a uniform 3 × 3 neuromorphic vision sensor array. Our work offers a platform to achieve motion detection based on controllable and energy-efficient ferroelectric programmability.

Polarization Pruning: Reliability Enhancement of Hafnia-Based Ferroelectric Devices for Memory and Neuromorphic Computing.

Ferroelectric (FE) materials are key to advancing electronic devices owing to their non-volatile properties, rapid state-switching abilities, and low-energy consumption. FE-based devices are used in logic circuits, memory-storage devices, sensors, and in-memory computing. However, the primary challenge in advancing the practical applications of FE-based memory is its reliability. To address this problem, a novel polarization pruning (PP) method is proposed. The PP is designed to eliminate weakly polarized domains by applying an opposite-sign pulse immediately after a program or erase operation. Significant improvements in the reliability of ferroelectric devices are achieved by reducing the depolarization caused by weakly polarized domains and mitigating the fluctuations in the ferroelectric dipole. These enhancements include a 25% improvement in retention, a 50% reduction in read noise, a 45% decrease in threshold voltage variation, and a 72% improvement in linearity. The proposed PP method significantly improves the memory storage efficiency and performance of neuromorphic systems.

Snn and sound: a comprehensive review of spiking neural networks in sound.

The rapid advancement of AI and machine learning has significantly enhanced sound and acoustic recognition technologies, moving beyond traditional models to more sophisticated neural network-based methods. Among these, Spiking Neural Networks (SNNs) are particularly noteworthy. SNNs mimic biological neurons and operate on principles similar to the human brain, using analog computing mechanisms. This capability allows for efficient sound processing with low power consumption and minimal latency, ideal for real-time applications in embedded systems. This paper reviews recent developments in SNNs for sound recognition, underscoring their potential to overcome the limitations of digital computing and suggesting directions for future research. The unique attributes of SNNs could lead to breakthroughs in mimicking human auditory processing more closely.

Electro-Optically Configurable Synaptic Transistors With Cluster-Induced Photoactive Dielectric Layer for Visual Simulation and Biomotor Stimuli.

The integration of visual simulation and biorehabilitation devices promises great applications for sustainable electronics, on-demand integration and neuroscience. However, achieving a multifunctional synergistic biomimetic system with tunable optoelectronic properties at the individual device level remains a challenge. Here, an electro-optically configurable transistor employing conjugated-polymer as semiconductor layer and an insulating polymer (poly(1,8-octanediol-co-citrate) (POC)) with clusterization-triggered photoactive properties as dielectric layer is shown. These devices realize adeptly transition from electrical to optical synapses, featuring multiwavelength and multilevel optical synaptic memory properties exceeding 3 bits. Utilizing enhanced optical memory, the images learning and memory function for visual simulation are achieved. Benefiting from rapid electrical response akin to biological muscle activation, increased actuation occurs under increased stimulus frequency of gate voltage. Additionally, the transistor on POC substrate can be effectively degraded in NaOH solution due to degradation of POC. Pioneeringly, the electro-optically configurability stems from light absorption and photoluminescence of the aggregation cluster in POC layer after 200 °C annealing. The enhancement of optical synaptic plasticity and integration of motion-activation functions within a single device opens new avenues at the intersection of optoelectronics, synaptic computing, and bioengineering.

Configurable Synaptic and Stochastic Neuronal Functions in ZnTe-Based Memristor for an RBM Neural Network.

This study presents findings that demonstrate the possibility of simplifying neural networks by inducing multifunctionality through separate manipulation within a single material. Herein, two-terminal memristor W/ZnTe/W devices implemented a multifunctional memristor comprising a selector, synapse, and a neuron using an ovonic threshold switching material. By setting the low-current level (µA) in the forming process, a stable memory-switching operation is achieved, and the capacity to implement a synapse is demonstrated based on paired-pulse facilitation/depression, potentiation/depression, spike-amplitude-dependent plasticity, and spike-number-dependent plasticity outcomes. Based on synaptic behavior, the Modified National Institute of Standards and Technology database image classification accuracy is up to 90%. Conversely, by setting the high-current level (mA) in the forming process, the stable bipolar threshold switching operation and good selector characteristics (300 ns switching speed, free-drift, recovery properties) are demonstrated. In addition, a stochastic neuron is implemented using the stochastic switching response in the positive voltage region. Utilizing stochastic neurons, it is possible to create a generative restricted Boltzmann machine model.

Energy Efficient Memristor Based on Green-Synthesized 2D Carbonyl-Decorated Organic Polymer and Application in Image Denoising and Edge Detection: Toward Sustainable AI.

According to the United Nations, around 53 million metric tons of electronic waste is produced every year, worldwide, the big majority of which goes unprocessed. With the rapid advances in AI technologies and adoption of smart gadgets, the demand for powerful logic and memory chips is expected to boom. Therefore, the development of green electronics is crucial to minimizing the impact of the alarmingly increasing e-waste. Here, it is shown the application of a green synthesized, chemically stable, carbonyl-decorated 2D organic, and biocompatible polymer as an active layer in a memristor device, sandwiched between potentially fully recyclable electrodes. The 2D polymer's ultramicro channels, decorated with C═O and O─H groups, efficiently promote the formation of copper nanofilaments. As a result, the device shows excellent bipolar resistive switching behavior with the potential to mimic synaptic plasticity. A large resistive switching window (103), low SET/RESET voltage of ≈0.5/-1.5 V), excellent device-to-device stability and synaptic features are demonstrated. Leveraging the device's synaptic characteristics, its applications in image denoising and edge detection is examined. The results show a reduction in power consumption by a factor of 103 compared to a traditional Tesla P40 graphics processing unit, indicating great promise for future sustainable AI-based applications.