Enabling an integrated rate-temporal learning scheme on memristor

Ultralow Energy Consumption and Fast Neuromorphic Computing Based on La0.1Bi0.9FeO3 Ferroelectric Tunnel Junctions

Low-power and fast artificial neural network devices represent the direction in developing analogue neural networks. Here, an ultralow power consumption (0.8 fJ) and rapid (100 ns) La0.1Bi0.9FeO3/La0.7Sr0.3MnO3 ferroelectric tunnel junction artificial synapse has been developed to emulate the biological neural networks. The visual memory and forgetting functionalities have been emulated based on long-term potentiation and depression with good linearity. Moreover, with a single device, logical operations of "AND" and "OR" are implemented, and an artificial neural network was constructed with a recognition accuracy of 96%. Especially for noisy data sets, the recognition speed is faster after preprocessing by the device in the present work. This sets the stage for highly reliable and repeatable unsupervised learning.

Volatile threshold switching and synaptic properties controlled by Ag diffusion using Schottky defects

We investigated diffusion memristors in the structure of Ag/Ta2O5/HfO2/Pt, in which active Ag ions control active metal ion diffusion and mimic biological brain functions. The CMOS compatible high-k metal oxide could control an Ag electrode that was ionized by applying an appropriate voltage to form a conductive filament, and the movement of Ag ions was chemically and electrically controlled due to oxygen density. This diffusion memristor exhibited diffused characteristics with a selectivity of 109, and achieved a low power consumption of 2 mW at a SET voltage of 0.2 V. The threshold transitions were reliably repeatable over 20 cycles for compliance currents of 10-6 A, 10-4 A, and no compliance current, with the largest standard deviation value of SET variation being 0.028. Upon filament formation, Ag ions readily diffused into the interface of the Ta2O5 and HfO2 layer, which was verified by investigating the Ag atomic percentage using XPS and vertical EDX and by measuring the relaxation time of 0.8 ms. Verified volatile switching device demonstrated the biological synaptic properties of quantum conductance, short-term memory, and long-term memory due to controlling the Ag. Diffusion memristors using designed control and switching layers as following film density and oxygen vacancy have improved results as low-power devices and neuromorphic devices compared to other diffusion-based devices, and these properties can be used for various applications such as selectors, synapses, and neuromorphic devices.

A Hybrid CMOS-Memristor Spiking Neural Network Supporting Multiple Learning Rules

Artificial intelligence (AI) is changing the way computing is performed to cope with real-world, ill-defined tasks for which traditional algorithms fail. AI requires significant memory access, thus running into the von Neumann bottleneck when implemented in standard computing platforms. In this respect, low-latency energy-efficient in-memory computing can be achieved by exploiting emerging memristive devices, given their ability to emulate synaptic plasticity, which provides a path to design large-scale brain-inspired spiking neural networks (SNNs). Several plasticity rules have been described in the brain and their coexistence in the same network largely expands the computational capabilities of a given circuit. In this work, starting from the electrical characterization and modeling of the memristor device, we propose a neuro-synaptic architecture that co-integrates in a unique platform with a single type of synaptic device to implement two distinct learning rules, namely, the spike-timing-dependent plasticity (STDP) and the Bienenstock-Cooper-Munro (BCM). This architecture, by exploiting the aforementioned learning rules, successfully addressed two different tasks of unsupervised learning.

Nanoelectronics Using Metal-Insulator Transition

Metal-insulator transition (MIT) coupled with an ultrafast, significant, and reversible resistive change in Mott insulators has attracted tremendous interest for investigation into next-generation electronic and optoelectronic devices, as well as a fundamental understanding of condensed matter systems. Although the mechanism of MIT in Mott insulators is still controversial, great efforts have been made to understand and modulate MIT behavior for various electronic and optoelectronic applications. In this review, recent progress in the field of nanoelectronics utilizing MIT is highlighted. A brief introduction to the physics of MIT and its underlying mechanisms is begun. After discussing the MIT behaviors of various Mott insulators, recent advances in the design and fabrication of nanoelectronics devices based on MIT, including memories, gas sensors, photodetectors, logic circuits, and artificial neural networks are described. Finally, an outlook on the development and future applications of nanoelectronics utilizing MIT is provided. This review can serve as an overview and a comprehensive understanding of the design of MIT-based nanoelectronics for future electronic and optoelectronic devices.

Perspective on Nanofluidic Memristors: from Mechanism to Application

Nanofluidic memristors are memory resistors based on nanoconfined fluidic systems exhibiting history-dependent ion conductivity. Toward establishing powerful computing systems beyond the Harvard architecture, these ion-based neuromorphic devices attracted enormous research attention owing to the unique characteristics of ion-based conductors. However, the design of nanofluidic memristor is still at a primary state and a systematic guidance on the rational design of nanofluidic system is desperately required for the development of nanofluidic-based neuromorphic devices. Herein, we proposed a systematic review on the history, main mechanism and potential application of nanofluidic memristors in order to give a prospective view on the design principle of memristors based on nanofluidic systems. Furthermore, based on the present status of these devices, some fundamental challenges for this promising area were further discussed to show the possible application of these ion-based devices.

Hardware Demonstration of SRDP Neuromorphic Computing with Online Unsupervised Learning Based on Memristor Synapses

Neuromorphic computing has shown great advantages towards cognitive tasks with high speed and remarkable energy efficiency. Memristor is considered as one of the most promising candidates for the electronic synapse of the neuromorphic computing system due to its scalability, power efficiency and capability to simulate biological behaviors. Several memristor-based hardware demonstrations have been explored to achieve the capacity of unsupervised learning with the spike-rate-dependent plasticity (SRDP) learning rule. However, the learning capacity is limited and few of the memristor-based hardware demonstrations have explored the online unsupervised learning at the network level with an SRDP algorithm. Here, we construct a memristor-based hardware system and demonstrate the online unsupervised learning of SRDP networks. The neuromorphic system consists of multiple memristor arrays as the synapse and the discrete CMOS circuit unit as the neuron. Unsupervised learning and online weight update of 10 MNIST handwritten digits are realized by the constructed SRDP networks, and the recognition accuracy is above 90% with 20% device variation. This work paves the way towards the realization of large-scale and efficient networks for more complex tasks.

Modeling-Based Design of Memristive Devices for Brain-Inspired Computing

Resistive switching random access memory (RRAM) has emerged for non-volatile memory application with the features of simple structure, low cost, high density, high speed, low power, and CMOS compatibility. In recent years, RRAM technology has made significant progress in brain-inspired computing paradigms by exploiting its unique physical characteristics, which attempts to eliminate the energy-intensive and time-consuming data transfer between the processing unit and the memory unit. The design of RRAM-based computing paradigms, however, requires a detailed description of the dominant physical effects correlated with the resistive switching processes to realize the interaction and optimization between devices and algorithms or architectures. This work provides an overview of the current progress on device-level resistive switching behaviors with detailed insights into the physical effects in the resistive switching layer and the multifunctional assistant layer. Then the circuit-level physics-based compact models will be reviewed in terms of typical binary RRAM and the emerging analog synaptic RRAM, which act as an interface between the device and circuit design. After that, the interaction between device and system performances will finally be addressed by reviewing the specific applications of brain-inspired computing systems including neuromorphic computing, in-memory logic, and stochastic computing.

Resistive Random Access Memory (RRAM): an Overview of Materials, Switching Mechanism, Performance, Multilevel Cell (mlc) Storage, Modeling, and Applications

In this manuscript, recent progress in the area of resistive random access memory (RRAM) technology which is considered one of the most standout emerging memory technologies owing to its high speed, low cost, enhanced storage density, potential applications in various fields, and excellent scalability is comprehensively reviewed. First, a brief overview of the field of emerging memory technologies is provided. The material properties, resistance switching mechanism, and electrical characteristics of RRAM are discussed. Also, various issues such as endurance, retention, uniformity, and the effect of operating temperature and random telegraph noise (RTN) are elaborated. A discussion on multilevel cell (MLC) storage capability of RRAM, which is attractive for achieving increased storage density and low cost is presented. Different operation schemes to achieve reliable MLC operation along with their physical mechanisms have been provided. In addition, an elaborate description of switching methodologies and current voltage relationships for various popular RRAM models is covered in this work. The prospective applications of RRAM to various fields such as security, neuromorphic computing, and non-volatile logic systems are addressed briefly. The present review article concludes with the discussion on the challenges and future prospects of the RRAM.

Extended memory lifetime in spiking neural networks employing memristive synapses with nonlinear conductance dynamics

Spiking neural networks (SNNs) employing memristive synapses are capable of life-long online learning. Because of their ability to process and classify large amounts of data in real-time using compact and low-power electronic systems, they promise a substantial technology breakthrough. However, the critical issue that memristor-based SNNs have to face is the fundamental limitation in their memory capacity due to finite resolution of the synaptic elements, which leads to the replacement of old memories with new ones and to a finite memory lifetime. In this study we demonstrate that the nonlinear conductance dynamics of memristive devices can be exploited to improve the memory lifetime of a network. The network is simulated on the basis of a spiking neuron model of mixed-signal digital-analogue sub-threshold neuromorphic CMOS circuits, and on memristive synapse models derived from the experimental nonlinear conductance dynamics of resistive memory devices when stimulated by trains of identical pulses. The network learning circuits implement a spike-based plasticity rule compatible with both spike-timing and rate-based learning rules. In order to get an insight on the memory lifetime of the network, we analyse the learning dynamics in the context of a classical benchmark of neural network learning, that is hand-written digit classification. In the proposed architecture, the memory lifetime and the performance of the network are improved for memristive synapses with nonlinear dynamics with respect to linear synapses with similar resolution. These results demonstrate the importance of following holistic approaches that combine the study of theoretical learning models with the development of neuromorphic CMOS SNNs with memristive devices used to implement life-long on-chip learning.

A multi-level memristor based on atomic layer deposition of iron oxide

This work reports the fabrication of memristive devices based on iron oxide (Fe₂O₃) thin films grown by atomic layer deposition (ALD) using ferrocene as iron precursor and ozone as oxidant. An excellent control of the ALD process was achieved by using an experimental procedure based on a sequence of micro-pulses, which provided long residence time and homogeneous diffusion of precursors, allowing ALD of thin films with smooth morphology and crystallinity which was found to increase with layer thickness, at temperatures as low as 250 °C. The resistive switching of symmetric Pt/Fe₂O₃/Pt thin film devices exhibited bipolar mode with good stability and endurance. Multi-level switching was achieved via current and voltage control. It was proved that the ON state regime can be tuned by changing the current compliance while the OFF state can be changed to intermediate levels by decreasing the maximum voltage during RESET. The structural analysis of the switched oxide layer revealed the presence of nano-sized crystalline domains corresponding to different iron oxide phases, suggesting that Joule heating effects during I-V cycling are responsible for a crystallization process of the pristine amorphous layer.

An On-Chip Trainable and the Clock-Less Spiking Neural Network With 1R Memristive Synapses

Spiking neural networks (SNNs) are being explored in an attempt to mimic brain's capability to learn and recognize at low power. Crossbar architecture with highly scalable resistive RAM or RRAM array serving as synaptic weights and neuronal drivers in the periphery is an attractive option for the SNN. Recognition (akin to "reading" the synaptic weight) requires small amplitude bias applied across the RRAM to minimize conductance change. Learning (akin to "writing" or updating the synaptic weight) requires large amplitude bias pulses to produce a conductance change. The contradictory bias amplitude requirement to perform reading and writing simultaneously and asynchronously, akin to biology, is a major challenge. Solutions suggested in the literature rely on time-division-multiplexing of read and write operations based on clocks, or approximations ignoring the reading when coincidental with writing. In this paper, we overcome this challenge and present a clock-less approach wherein reading and writing are performed in different frequency domains. This enables learning and recognition simultaneously on an SNN. We validate our scheme in SPICE circuit simulator by translating a two-layered feed-forward Iris classifying SNN to demonstrate software-equivalent performance. The system performance is not adversely affected by a voltage dependence of conductance in realistic RRAMs, despite departing from linearity. Overall, our approach enables direct implementation of biological SNN algorithms in hardware.

Artificial Synapses Emulated by an Electrolyte-Gated Tungsten-Oxide Transistor

Considering that the human brain uses ≈1015 synapses to operate, the development of effective artificial synapses is essential to build brain-inspired computing systems. In biological synapses, the voltage-gated ion channels are very important for regulating the action-potential firing. Here, an electrolyte-gated transistor using WO3 with a unique tunnel structure, which can emulate the ionic modulation process of biological synapses, is proposed. The transistor successfully realizes synaptic functions of both short-term and long-term plasticity. Short-term plasticity is mimicked with the help of electrolyte ion dynamics under low electrical bias, whereas the long-term plasticity is realized using proton insertion in WO3 under high electrical bias. This is a new working approach to control the transition from short-term memory to long-term memory using different gate voltage amplitude for artificial synapses. Other essential synaptic behaviors, such as paired pulse facilitation, the depression and potentiation of synaptic weight, as well as spike-timing-dependent plasticity are also implemented in this artificial synapse. These results provide a new recipe for designing synaptic electrolyte-gated transistors through the electrostatic and electrochemical effects.

Memristive Devices with Highly Repeatable Analog States Boosted by Graphene Quantum Dots

Memristive devices, having a huge potential as artificial synapses for low-power neural networks, have received tremendous attention recently. Despite great achievements in demonstration of plasticity and learning functions, little progress has been made in the repeatable analog resistance states of memristive devices, which is, however, crucial for achieving controllable synaptic behavior. The controllable behavior of synapse is highly desired in building neural networks as it helps reduce training epochs and diminish error probability. Fundamentally, the poor repeatability of analog resistance states is closely associated with the random formation of conductive filaments, which consists of oxygen vacancies. In this work, graphene quantum dots (GQDs) are introduced into memristive devices. By virtue of the abundant oxygen anions released from GQDs, the GQDs can serve as nano oxygen-reservoirs and enhance the localization of filament formation. As a result, analog resistance states with highly tight distribution are achieved with nearly 85% reduction in variations. In addition the insertion of GQDs can alter the energy band alignment and boost the tunneling current, which leads to significant reduction in both switching voltages and their distribution variations. This work may pave the way for achieving artificial neural networks with accurate and efficient learning capability.

Coupling Resistive Switching Devices with Neurons: State of the Art and Perspectives

Here we provide the state-of-the-art of bioelectronic interfacing between biological neuronal systems and artificial components, focusing the attention on the potentiality offered by intrinsically neuromorphic synthetic devices based on Resistive Switching (RS). Neuromorphic engineering is outside the scopes of this Perspective. Instead, our focus is on those materials and devices featuring genuine physical effects that could be sought as non-linearity, plasticity, excitation, and extinction which could be directly and more naturally coupled with living biological systems. In view of important applications, such as prosthetics and future life augmentation, a cybernetic parallelism is traced, between biological and artificial systems. We will discuss how such intrinsic features could reduce the complexity of conditioning networks for a more natural direct connection between biological and synthetic worlds. Putting together living systems with RS devices could represent a feasible though innovative perspective for the future of bionics.

Hierarchical Chunking of Sequential Memory on Neuromorphic Architecture with Reduced Synaptic Plasticity

Chunking refers to a phenomenon whereby individuals group items together when performing a memory task to improve the performance of sequential memory. In this work, we build a bio-plausible hierarchical chunking of sequential memory (HCSM) model to explain why such improvement happens. We address this issue by linking hierarchical chunking with synaptic plasticity and neuromorphic engineering. We uncover that a chunking mechanism reduces the requirements of synaptic plasticity since it allows applying synapses with narrow dynamic range and low precision to perform a memory task. We validate a hardware version of the model through simulation, based on measured memristor behavior with narrow dynamic range in neuromorphic circuits, which reveals how chunking works and what role it plays in encoding sequential memory. Our work deepens the understanding of sequential memory and enables incorporating it for the investigation of the brain-inspired computing on neuromorphic architecture.

Nonassociative learning implementation by a single memristor-based multi-terminal synaptic device

Animals' survival is dependent on their abilities to adapt to the changing environment by adjusting their behaviours, which is related to the ubiquitous learning behaviour, nonassociative learning. Thus mimicking the indispensable learning behaviour in organisms based on electronic devices is vital to better achieve artificial neural networks and neuromorphic computing. Here a three terminal device consisting of an oxide-based memristor and a NMOS transistor is proposed. The memristor with gradual conductance tuning inherently functions as the synapse between sensor neurons and motor neurons and presents adjustable synaptic plasticity, while the NMOS transistor attached to the memristor is utilized to mimic the modulatory effect of the neuromodulator released by inter neurons. Such a memristor-based multi-terminal device allows the practical implementation of significant nonassociative learning based on a single electronic device. In this study, the experience-induced modification behaviour, both habituation and sensitization, was successfully achieved. The dependence of the nonassociative behavioural response on the strength and interval of presented stimuli was also discussed. The implementation of nonassociative learning offers feasible and experimental advantages for further study on neuromorphic systems based on electronic devices.

Long-Term Homeostatic Properties Complementary to Hebbian Rules in CuPc-Based Multifunctional Memristor

Most simulations of neuroplasticity in memristors, which are potentially used to develop artificial synapses, are confined to the basic biological Hebbian rules. However, the simplex rules potentially can induce excessive excitation/inhibition, even collapse of neural activities, because they neglect the properties of long-term homeostasis involved in the frameworks of realistic neural networks. Here, we develop organic CuPc-based memristors of which excitatory and inhibitory conductivities can implement both Hebbian rules and homeostatic plasticity, complementary to Hebbian patterns and conductive to the long-term homeostasis. In another adaptive situation for homeostasis, in thicker samples, the overall excitement under periodic moderate stimuli tends to decrease and be recovered under intense inputs. Interestingly, the prototypes can be equipped with bio-inspired habituation and sensitization functions outperforming the conventional simplified algorithms. They mutually regulate each other to obtain the homeostasis. Therefore, we develop a novel versatile memristor with advanced synaptic homeostasis for comprehensive neural functions.

Unsupervised learning in probabilistic neural networks with multi-state metal-oxide memristive synapses

In an increasingly data-rich world the need for developing computing systems that cannot only process, but ideally also interpret big data is becoming continuously more pressing. Brain-inspired concepts have shown great promise towards addressing this need. Here we demonstrate unsupervised learning in a probabilistic neural network that utilizes metal-oxide memristive devices as multi-state synapses. Our approach can be exploited for processing unlabelled data and can adapt to time-varying clusters that underlie incoming data by supporting the capability of reversible unsupervised learning. The potential of this work is showcased through the demonstration of successful learning in the presence of corrupted input data and probabilistic neurons, thus paving the way towards robust big-data processors.

Investigation and Manipulation of Different Analog Behaviors of Memristor as Electronic Synapse for Neuromorphic Applications

Low-power and high-density electronic synapse is an important building block of brain-inspired systems. The recent advancement in memristor has provided an opportunity to advance electronic synapse design. However, a guideline on designing and manipulating the memristor’s analog behaviors is still lacking. In this work, we reveal that compliance current (I_comp) of electroforming process played an important role in realizing a stable analog behavior, which is attributed to the generation of conical-type conductive filament. A proper I_comp could result in a large conductance window, good stability, and low voltage analog switching. We further reveal that different pulse conditions can lead to three analog behaviors, where the conductance changes in monotonic increase, plateau after initial jump, and impulse-like shape, respectively. These behaviors could benefit the design of electronic synapse with enriched learning capabilities. This work will provide a useful guideline for designing and manipulating memristor as electronic synapses for brain-inspired systems.

Synaptic Metaplasticity Realized in Oxide Memristive Devices

Metaplasticity, a higher order of synaptic plasticity, as well as a key issue in neuroscience, is realized with artificial synapses based on a WO3 thin film, and the activity-dependent metaplastic responses of the artificial synapses, such as spike-timing-dependent plasticity, are systematically investigated. This work has significant implications in neuromorphic computation.

Simulation of synaptic short-term plasticity using Ba(CF3SO3)2-doped polyethylene oxide electrolyte film

The simulation of synaptic plasticity using new materials is critical in the study of brain-inspired computing. Devices composed of Ba(CF3SO3)2-doped polyethylene oxide (PEO) electrolyte film were fabricated and with pulse responses found to resemble the synaptic short-term plasticity (STP) of both short-term depression (STD) and short-term facilitation (STF) synapses. The values of the charge and discharge peaks of the pulse responses did not vary with input number when the pulse frequency was sufficiently low(~1 Hz). However, when the frequency was increased, the charge and discharge peaks decreased and increased, respectively, in gradual trends and approached stable values with respect to the input number. These stable values varied with the input frequency, which resulted in the depressed and potentiated weight modifications of the charge and discharge peaks, respectively. These electrical properties simulated the high and low band-pass filtering effects of STD and STF, respectively. The simulations were consistent with biological results and the corresponding biological parameters were successfully extracted. The study verified the feasibility of using organic electrolytes to mimic STP.

Single-Readout High-Density Memristor Crossbar

High-density memristor-crossbar architecture is a very promising technology for future computing systems. The simplicity of the gateless-crossbar structure is both its principal advantage and the source of undesired sneak-paths of current. This parasitic current could consume an enormous amount of energy and ruin the readout process. We introduce new adaptive-threshold readout techniques that utilize the locality and hierarchy properties of the computer-memory system to address the sneak-paths problem. The proposed methods require a single memory access per pixel for an array readout. Besides, the memristive crossbar consumes an order of magnitude less power than state-of-the-art readout techniques.

Frequency-dependent learning achieved using semiconducting polymer/electrolyte composite cells

Frequency-dependent learning has been achieved using semiconducting polymer/electrolyte composite cells. The cells composed of polymer/electrolyte double layers realized the conventional spike-rate-dependent plasticity (SRDP) learning model. These cells responded to depression upon low-frequency stimulation and to potentiation upon high-frequency stimulation and presented long-term memory. The transition threshold θm from depression to potentiation varied depending on the previous stimulations. A nanostructure resembling a bio-synapse in its transport passages was demonstrated and a random channel model was proposed to describe the ionic kinetics at the polymer/electrolyte interface during and after stimulations with various frequencies, accounting for the observed SRDP.

Complex Learning in Bio-plausible Memristive Networks

The emerging memristor-based neuromorphic engineering promises an efficient computing paradigm. However, the lack of both internal dynamics in the previous feedforward memristive networks and efficient learning algorithms in recurrent networks, fundamentally limits the learning ability of existing systems. In this work, we propose a framework to support complex learning functions by introducing dedicated learning algorithms to a bio-plausible recurrent memristive network with internal dynamics. We fabricate iron oxide memristor-based synapses, with well controllable plasticity and a wide dynamic range of excitatory/inhibitory connection weights, to build the network. To adaptively modify the synaptic weights, the comprehensive recursive least-squares (RLS) learning algorithm is introduced. Based on the proposed framework, the learning of various timing patterns and a complex spatiotemporal pattern of human motor is demonstrated. This work paves a new way to explore the brain-inspired complex learning in neuromorphic systems.

Effect of heavy-ion on frequency selectivity of semiconducting polymer/electrolyte heterojunction

Long-term bidirectional frequency selectivity has been achieved in MEH-PPV/PEO–Nd³⁺cells, which suggests spike-rate-dependent plasticity learning protocol. It depends on pulse shape due to variation of ionic type.