Dynamic Memristor Research Articles

Convolutional neural network for high-performance reservoir computing using dynamic memristors

In the rapidly advancing field of neuromorphic computing, W/ZnO/TiN resistive random-access memory (RRAM) devices have emerged as a next-generation computational building block. Our findings reveal the significant role played by the thickness of the ZnO layer in determining the electrical properties essential for data storage and neuromorphic applications. The short-term memory (STM) capabilities, which are critical for processing temporal information, are closely examined alongside their potential to simulate biological synaptic functions through multilevel conductance states and synaptic behaviors such as paired-pulse facilitation. Integrating these devices into reservoir computing systems enhances pattern recognition and accelerates learning, which demonstrates their utility in sequential data processing. In addition, conductance modulation via pulse width adjustment is a novel strategy to optimize memory device performance. By showcasing the effectiveness of W/ZnO/TiN devices in neuromorphic computing through high-accuracy image recognition tasks, our study highlights their foundational role in advancing neuromorphic computing technologies. The adaptability, learning capabilities, and efficiency of these devices underscore their potential for developing hardware-based neuromorphic systems that are capable of complex data processing.

Evaporated Copper‐Based Perovskite Dynamic Memristors for Reservoir Computing Systems

AbstractDynamic memristors are considered as the optimal hardware devices for reservoir computing (RC) enabled by their nonlinear conductance variations. This significantly reduces the extensive training workload typically required by traditional neural networks. Lead halide perovskites, with their tunable band structure and active ion migration properties, have emerged as highly promising materials for developing dynamic memristors. However, large‐scale and consistently stable production remains a challenge for perovskite functional films, while lead elements' toxicity and environmental impact also partly restrict their practical device utilization. In this work, lead‐free copper‐based perovskite (i.e., CsCu2I3) films are prepared by thermal evaporation for constructing dynamic memristors. The effective conductivity modulation of CsCu2I3‐based memristor can be utilized in artificial neural networks, achieving a high handwritten digit recognition accuracy of 91.2%. In addition, the RC system is also constructed based on the dynamic behavior of the devices, by which a letter recognition accuracy of 98.2% with simple training is achieved. This technology provides a feasible pathway to construct copper‐based perovskite dynamic memristors for future neural network information processing.

Enhancing large-area Geiger-mode avalanche photodiode performance through dynamic memristor quenching: a study on improving count rate, reducing jitter and mitigating afterpulsing

While the larger photosensitive area of Geiger-mode avalanche photodiodes (GmAPDs) enhances their detection range and signal collection, improving their utility in weak light detection, their practicality is limited by a long recovery time, high afterpulsing probability (AP) and excessive jitter. Utilizing a dynamic memristor as a quenching resistor, this research improves the count rate of a large-size GmAPD by 100× at an overvoltage of 2.5 V, compared with a fixed resistor-quenched GmAPD. Furthermore, at a photon pulse frequency of 1 MHz jitter time is reduced from 3.60 ns to 0.48 ns, and the afterpulsing probability is effectively mitigated from 30.88% to 8.58%.

Temporal multibit operation of dynamic memristor for reservoir computing

The exponential growth of data in our society following the fourth industrial revolution has exposed the limitations of existing technologies. The current computing architecture struggles to process efficiently the immense volume of data generated every second. A potential breakthrough to overcome these limitations involves the adoption of neuromorphic computing to emulate the functionalities of a biological brain. In this context, a synaptic diode composed of a Pt/TiOx/Ti stack is proposed to replicate the operations of synapses and neurons. Various synapse function emulations, including potentiation, depression, and spike-rate-dependent plasticity, are examined to completely realize an artificial synapse. Additionally, precise control over pulse sequences is implemented for cost-effective and high-density reservoir computing, exploring configurations with 2, 4, and 7 bits. Thus, showcasing the synaptic diode capable of diverse functions with computing adaptability, enabling favorites in future applications of neuromorphic computing.

Nonmasking-based reservoir computing with a single dynamic memristor for image recognition

Nonmasking-based reservoir computing with a single dynamic memristor for image recognition

Fully Memristive Elementary Motion Detectors for a Maneuver Prediction.

Insects can efficiently perform object motion detection via a specialized neural circuit, called an elementary motion detector (EMD). In contrast, conventional machine vision systems require significant computational resources for dynamic motion processing. Here, a fully memristive EMD (M-EMD) is presented that implements the Hassenstein-Reichardt (HR) correlator, a biological model of the EMD. The M-EMD consists of a simple Wye (Y) configuration, including a static resistor, a dynamic memristor, and a Mott memristor. The resistor and dynamic memristor introduce different signal delays, enabling spatio-temporal signal integration in the subsequent Mott memristor, resulting in a direction-selective response. In addition, a neuromorphic system is developed employing the M-EMDs to predict a lane-changing maneuver by vehicles on the road. The system achieved a high accuracy (> 87%) in predicting future lane-changing maneuvers on the Next Generation Simulation (NGSIM) dataset while reducing the computational cost by 92.9% compared to the conventional neuromorphic system without the M-EMD, suggesting its strong potential for edge-level computing.

Brain-Inspired Reservoir Computing Using Memristors with Tunable Dynamics and Short-Term Plasticity.

Recent advancements in reservoir computing (RC) research have created a demand for analogue devices with dynamics that can facilitate the physical implementation of reservoirs, promising faster information processing while consuming less energy and occupying a smaller area footprint. Studies have demonstrated that dynamic memristors, with nonlinear and short-term memory dynamics, are excellent candidates as information-processing devices or reservoirs for temporal classification and prediction tasks. Previous implementations relied on nominally identical memristors that applied the same nonlinear transformation to the input data, which is not enough to achieve a rich state space. To address this limitation, researchers either diversified the data encoding across multiple memristors or harnessed the stochastic device-to-device variability among the memristors. However, this approach requires additional preprocessing steps and leads to synchronization issues. Instead, it is preferable to encode the data once and pass them through a reservoir layer consisting of memristors with distinct dynamics. Here, we demonstrate that ion-channel-based memristors with voltage-dependent dynamics can be controllably and predictively tuned through the voltage or adjustment of the ion channel concentration to exhibit diverse dynamic properties. We show, through experiments and simulations, that reservoir layers constructed with a small number of distinct memristors exhibit significantly higher predictive and classification accuracies with a single data encoding. We found that for a second-order nonlinear dynamical system prediction task, the varied memristor reservoir experimentally achieved an impressive normalized mean square error of 1.5 × 10-3, using only five distinct memristors. Moreover, in a neural activity classification task, a reservoir of just three distinct memristors experimentally attained an accuracy of 96.5%. This work lays the foundation for next-generation physical RC systems that can exploit the complex dynamics of their diverse building blocks to achieve increased signal processing capabilities.

Dynamic memristor array with multiple reservoir states for training efficient neuromorphic computing

Exploration of efficient neuromorphic computing using Pt/Al/TiOy/TiOx/Al2O3/Pt array memristors implemented a reservoir with 16 states, demonstrating the training process of synaptic array images.

Dynamic memristor for physical reservoir computing.

Reservoir computing (RC) has attracted considerable attention for its efficient handling of temporal signals and lower training costs. As a nonlinear dynamic system, RC can map low-dimensional inputs into high-dimensional spaces and implement classification using a simple linear readout layer. The memristor exhibits complex dynamic characteristics due to its internal physical processes, which renders them an ideal choice for the implementation of physical reservoir computing (PRC) systems. This review focuses on PRC systems based on memristors, explaining the resistive switching mechanism at the device level and emphasizing the tunability of their dynamic behavior. The development of memristor-based reservoir computing systems is highlighted, along with discussions on the challenges faced by this field and potential future research directions.

Unveiling transient current response in bilayer oxide-based physical reservoirs for time-series data analysis.

Physical reservoirs employed to map time-series data and analyze extracted features have attracted interest owing to their low training cost and mitigated interconnection complexity. This study reports a physical reservoir based on a bilayer oxide-based dynamic memristor. The proposed device exhibits a nonlinear current response and short-term memory (STM), satisfying the requirements of reservoir computing (RC). These characteristics are validated using a compact model to account for resistive switching (RS) via the dynamic evolution of the internal state variable and the relocation of oxygen vacancies. Mathematically, the transient current response can be quantitatively described according to a simple set of equations to correlate the theoretical framework with experimental results. Furthermore, the device shows significant reliability and ability to distinguish 4-bit inputs and four diverse neural firing patterns. Therefore, this work shows the feasibility of implementing physical reservoirs in hardware and advances the understanding of the dynamic response.

IGZO/SnOx-based dynamic memristor with fading memory effect for reservoir computing

We investigate a synaptic device with short-term memory characteristics using IGZO/SnOx as the switching layer. The thickness and components of each layer are analyzed by using x-ray photoelectron spectroscopy and transmission electron microscopy. The memristor exhibits analog resistive switching and a volatile feature with current decay over time. Moreover, through ten cycles of potentiation and depression, we demonstrate stable conductance modulation, leading to high-accuracy Modified National Institute of Standards and Technology pattern recognition. We effectively emulate the learning system of a biological synapse, including paired-pulse facilitation, spiking-amplitude-dependent plasticity, and spiking-rate-dependent plasticity (SRDP) by pulse trains. Ultimately, 4-bit reservoir computing divided into 16 states is incarnated using a pulse stream considering short-term memory plasticity and decay properties.

Artificial synapses based on 2D-layered palladium diselenide heterostructure dynamic memristor for neuromorphic applications

The transformation of partially amorphous two-dimensional (2D) material layers represents a promising avenue for enhancing the reliability of heterostructure memristor devices and achieving high-density memory with synaptic characteristics. In this study, we investigate the advantages of aluminum oxide-based redox reactive layers integrated with 2D pentagonal palladium diselenide (PdSe2) layers, resulting in the realization of a reliable multilevel conductance memory controlled by DC and pulse voltages. Through careful analysis, we observed that the formation and rupture of conductive filaments related to oxygen vacancies were significantly controlled at the Al2O3/PdSe2 interface, where a mixed oxide of Al-PdSeOx was formed. This exciting phenomenon was further corroborated by comprehensive chemical analysis of the synapse structure as well as various synaptic behaviors, predominantly modulated by the facile movement of oxygen ions under the influence of external electric fields. Remarkably, our experimental results demonstrated improved endurance with multiple memory states achieved by judiciously altering the RESET voltage and SET current compliance, yielding an impressive memory window exceeding 10. Moreover, we successfully achieve a controlled transition from short-term to long-term memory through the application of identical and nonidentical consecutive pulse voltages. Inspired by biological synapses, our PdSe2-based artificial synapse exhibits biocompatible short-term plasticity, such as spike rate-dependent plasticity, paired-pulse facilitation, and experience-dependent plasticity, effectively mimicking the functionality of its biological counterparts. Notably, this biocompatible synapse also demonstrates great potential for visual memory processing, and multilevel 4-bit reservoir computing by demonstrating 5 × 4 digit image recognition thereby offering a distinct advantage for artificial neuromorphic applications.

Seizure detection using dynamic memristor-based reservoir computing and leaky integrate-and-fire neuron for post-processing

Epilepsy is a prevalent neurological disorder, rendering the development of automated seizure detection systems imperative. While complex machine learning models are powerful, their training and hardware deployment remain challenging. The reservoir computing system offers a low-cost solution in terms of both hardware requirements and training. In this paper, we introduce a compact reservoir computing system for seizure detection, based on the α-In2Se3 dynamic memristors. Leaky integrate-and-fire neurons are used for post-processing the output of the system, and experimental results indicate their effectiveness in suppressing erroneous outputs, where both accuracy and specificity are enhanced by over 2.5%. The optimized compact reservoir system achieves 96.40% accuracy, 86.34% sensitivity, and 96.56% specificity in seizure detection tasks. This work demonstrates the feasibility of using reservoir computing for seizure detection and shows its potential for future application in extreme edge devices.

Generative complex networks within a dynamic memristor with intrinsic variability

Artificial neural networks (ANNs) have gained considerable momentum in the past decade. Although at first the main task of the ANN paradigm was to tune the connection weights in fixed-architecture networks, there has recently been growing interest in evolving network architectures toward the goal of creating artificial general intelligence. Lagging behind this trend, current ANN hardware struggles for a balance between flexibility and efficiency but cannot achieve both. Here, we report on a novel approach for the on-demand generation of complex networks within a single memristor where multiple virtual nodes are created by time multiplexing and the non-trivial topological features, such as small-worldness, are generated by exploiting device dynamics with intrinsic cycle-to-cycle variability. When used for reservoir computing, memristive complex networks can achieve a noticeable increase in memory capacity a and respectable performance boost compared to conventional reservoirs trivially implemented as fully connected networks. This work expands the functionality of memristors for ANN computing.

LiNbO3 dynamic memristors for reservoir computing.

Information in conventional digital computing platforms is encoded in the steady states of transistors and processed in a quasi-static way. Memristors are a class of emerging devices that naturally embody dynamics through their internal electrophyiscal processes, enabling nonconventional computing paradigms with enhanced capability and energy efficiency, such as reservoir computing. Here, we report on a dynamic memristor based on LiNbO3. The device has nonlinear I-V characteristics and exhibits short-term memory, suitable for application in reservoir computing. By time multiplexing, a single device can serve as a reservoir with rich dynamics which used to require a large number of interconnected nodes. The collective states of five memristors after the application of trains of pulses to the respective memristors are unique for each combination of pulse patterns, which is suitable for sequence data classification, as demonstrated in a 5 × 4 digit image recognition task. This work broadens the spectrum of memristive materials for neuromorphic computing.

Multistability of Dynamic Memristor Delayed Cellular Neural Networks With Application to Associative Memories.

Recently, dynamic memristor (DM)-cellular neural networks (CNNs) have received widespread attention due to their advantage of low power consumption. The previous works showed that DM-CNNs have at most 318 equilibrium points (EPs) with n=16 cells. Since time delay is unavoidable during the process of information transmission, the goal of this article is to research the multistability of DM-CNNs with time delay, and, meanwhile, to increase the storage capacity of DM-delay (D)CNNs. Depending on the different constitutive relations of memristors, two cases of the multistability for DM-DCNNs are discussed. After determining the constitutive relations, the number of EPs of DM-DCNNs is increased to 3n with n cells by means of the appropriate state-space decomposition and the Brouwer's fixed point theorem. Furthermore, the enlarged attraction domains of EPs can be obtained, and 2n of these EPs are locally exponentially stable in two cases. Compared with standard CNNs, the dynamic behavior of DM-DCNNs shows an outstanding merit. That is, the value of voltage and current approach to zero when the system becomes stable, and the memristor provides a nonvolatile memory to store the computation results. Finally, two numerical simulations are presented to illustrate the effectiveness of the theoretical results, and the applications of associative memories are shown at the end of this article.

Adaptive Memory-Enhanced Time Delay Reservoir and its Memristive Implementation

Time Delay Reservoir (TDR) is a hardware-friendly machine learning approach from two perspectives. First, it can prevent the connection overhead of neural networks with increasing neurons. Second, through its dynamic system representation, TDR can also be implemented in hardware by different systems. However, it performs poorly on tasks that involve long-term dependency. In this work, we first introduce a higher-order delay unit, which is capable of accumulating and transferring the long history states in an adaptive manner to further enhance the reservoir memory. Particle Swarm Optimisation is applied to optimize the enhanced degree of memory adaptivity. Our experiments demonstrate its superiority both for short- and long-term memory datasets over seven existing approaches. In light of the hardware-friendly feature of TDR, we further propose a memristive implementation of our adaptive memory-enhanced TDR, where a dynamic memristor and the memristor-based delay element are applied to construct the reservoir. Through circuit simulation, the feasibility of our proposed memristive implementation is verified. The comparisons with different hardware reservoirs show that our proposed memristive implementation is effective both for short- and long-term memory datasets, while exhibiting benefits in terms of smaller circuit area and lower power consumption compared with traditional hardware reservoirs.

A memristor-based analogue reservoir computing system for real-time and power-efficient signal processing

Reservoir computing offers a powerful neuromorphic computing architecture for spatiotemporal signal processing. To boost the power efficiency of the hardware implementations of reservoir computing systems, analogue devices and components—including spintronic oscillators, photonic modules, nanowire networks and memristors—have been used to partially replace the elements of fully digital systems. However, the development of fully analogue reservoir computing systems remains limited. Here we report a fully analogue reservoir computing system that uses dynamic memristors for the reservoir layer and non-volatile memristors for the readout layer. The system can efficiently process spatiotemporal signals in real time with three orders of magnitude lower power consumption than digital hardware. We illustrate the capabilities of the system using temporal arrhythmia detection and spatiotemporal dynamic gesture recognition tasks, achieving accuracies of 96.6% and 97.9%, respectively. Our memristor-based fully analogue reservoir computing system could be of use in edge computing applications that require extremely low power and hardware cost.

A Memristors-Based Dendritic Neuronfor High-Efficiency Spatial-Temporal Information Processing.

Diverse microscopic ionic dynamics help mediate the ability of a biological neural network to handle complex tasks with low energy consumption. Thus, rich internal ionic dynamics in memristors based on transition metal oxide are expected to provide a unique and useful platform for implementing energy-efficient neuromorphic computing. To this end, a titanium oxide (TiOx )-based interface-type dynamic memristor and an niobium oxide (NbOx )-based Mott memristor are integrated as an artificial dendrite and spike-firing soma, respectively, to construct a dendritic neuron unit for realizing high-efficiency spatial-temporal information processing. Further, a dendritic neural network is hardware-implemented for spatial-temporal information processing to highlight the computational advantages achieved by incorporating dendritic functions in the network. Human motion recognition is demonstrated using the Nanyang Technological University-Red Green Blue(NTU-RGB) dataset as a benchmark spatial-temporal task; it shows a nearly 20% improvement in accuracy for the memristors-based hardware incorporating dendrites and a 1000× advantage in power efficiency compared to that of the graphics processing unit (GPU). The dendritic neuron developed in this study can be considered a critical building block for implementing more bio-plausible neural networks that can manage complex spatial-temporal tasks with high efficiency.

Experimental demonstration of highly reliable dynamic memristor for artificial neuron and neuromorphic computing

Neuromorphic computing, a computing paradigm inspired by the human brain, enables energy-efficient and fast artificial neural networks. To process information, neuromorphic computing directly mimics the operation of biological neurons in a human brain. To effectively imitate biological neurons with electrical devices, memristor-based artificial neurons attract attention because of their simple structure, energy efficiency, and excellent scalability. However, memristor’s non-reliability issues have been one of the main obstacles for the development of memristor-based artificial neurons and neuromorphic computings. Here, we show a memristor 1R cross-bar array without transistor devices for individual memristor access with low variation, 100% yield, large dynamic range, and fast speed for artificial neuron and neuromorphic computing. Based on the developed memristor, we experimentally demonstrate a memristor-based neuron with leaky-integrate and fire property with excellent reliability. Furthermore, we develop a neuro-memristive computing system based on the short-term memory effect of the developed memristor for efficient processing of sequential data. Our neuro-memristive computing system successfully trains and generates bio-medical sequential data (antimicrobial peptides) while using a small number of training parameters. Our results open up the possibility of memristor-based artificial neurons and neuromorphic computing systems, which are essential for energy-efficient edge computing devices.