Spike-rate-dependent Plasticity Research Articles

Synaptic Plasticity and Metaplasticity of Biological Synapse Realized in a KNbO3 Memristor for Application to Artificial Synapse.

Amorphous KNbO3 (KN) films were grown on a TiN/SiO2/Si substrate to synthesize a KN memristor as a potential artificial synapse. The Pt/KN/TiN memristor exhibited typical and reliable bipolar switching behavior with multiple resistance levels. It also showed the transmission properties of a biological synapse, with a good conductance modulation linearity. Moreover, the KN memristor can emulate various biological synaptic plasticity characteristics including short-term plasticity, long-term plasticity, spike-rate dependent plasticity, paired-pulse facilitation, and post-tetanic potentiation by controlling the number and rate of the potentiation spike. Spike-timing-dependent plasticity (STDP), which is an essential property of biological synapses, is also realized in the KN memristor. The synaptic plasticity of the KN memristor can be explained by oxygen vacancy movement and oxygen vacancy filaments. The metaplasticity of biological synapses was also implemented in the KN memristor, including the metaplasticity of long-term potentiation and depression, and of STDP. Therefore, the KN memristor could be used as an artificial synapse in neuromorphic computing systems.

Structural engineering of tantalum oxide based memristor and its electrical switching responses using rapid thermal annealing

We have demonstrated the co-existence of reliable analog and digital switching characteristics with tantalum oxide based memristor by appropriate rapid thermal annealing (RTA). The device without RTA exhibits a digital SET and multilevel RESET for positive and negative sweeps, respectively. On the other hands, the device shows only analog switching characteristics such that current level increases and decreases gradually for successive positive and negative voltage sweeps, respectively, before any electroforming process with the RTA in the nitrogen ambient at the crystalline temperature of tantalum oxide which is 700 °C for 60 s. Once electroforming process is done, the device exhibits a reliable digital switching with SET at positive sweep and RESET at negative sweep voltages. In the analog state of the device we successfully emulate the synaptic characteristic of the device like spike-rate dependent plasticity (SRDP), pulse-paired facilitation (PPF) and post-tetanic potentiation (PTP). Finally, the Hermann Ebbinghaus forgetting curve is obtained from these devices. The conversion of the device from the digital SET and multilevel RESET to analogue switching is attributed to structural transition of amorphous tantalum oxide to polycrystalline tantalum oxide, different defect density and interface variation in the device.

Mimicking synaptic plasticity and learning behaviours in solution processed SnO2 memristor

In this study, a transparent memristor with a configuration of Au/SnO2/FTO is fabricated by a simple solution process at low temperature and further utilized to mimic biological synapses. A series of significant synaptic functions, including nonlinear transmission characteristics, spike-rate-dependent plasticity (SRDP), short-term plasticity (STP) and long-term plasticity (LTP) are emulated. The transition from short-term to long-term plasticity is also investigated in the device by repeated stimulation. The nonlinear rectification characteristic in the current memristor is attributed to the Schottky barrier at the Au/SnO2 interface. By controlling the oxygen vacancy migration induced under electrical input, the barrier at the interface can be modified, giving rise to the different synaptic functions. These results suggest that the proposed Au/SnO2/FTO memristor in this study is a promising synaptic device for artificial neural network applications.

Long-Term Synaptic Plasticity Emulated in Modified Graphene Oxide Electrolyte Gated IZO-Based Thin-Film Transistors.

Emulating neural behaviors at the synaptic level is of great significance for building neuromorphic computational systems and realizing artificial intelligence. Here, oxide-based electric double-layer (EDL) thin-film transistors were fabricated using 3-triethoxysilylpropylamine modified graphene oxide (KH550-GO) electrolyte as the gate dielectrics. Resulting from the EDL effect and electrochemical doping between mobile protons and the indium-zinc-oxide channel layer, long-term synaptic plasticity was emulated in our devices. Synaptic functions including long-term memory, synaptic temporal integration, and dynamic filters were successfully reproduced. In particular, spike rate-dependent plasticity (SRDP), one of the basic learning rules of long-term plasticity in the neural network where the synaptic weight changes according to the rate of presynaptic spikes, was emulated in our devices. Our results may facilitate the development of neuromorphic computational systems.

Implementation of Memristive Neural Network With Full-Function Pavlov Associative Memory

In this paper, implementation of memristive neural network with full-function Pavlov associative memory is designed based on a proposed associative memory rule. The designed neural network can well perform the Pavlov associative memory in the network with at least three interconnected neurons. This neural network and the associative memory rule that is partly based on spike-rate-dependent plasticity (SRDP) protocol are inspired by the famous Pavlov’s dog-experiment that demonstrated the interrelation between the “sight of food” and the “ringing.” Besides the learning activity, the proposed network can also perform two kinds of forgetting activities after the learning process is completed: on one hand, when the salivation neuron is stimulated by the food neuron alone, after a period of time, the ring neuron can no longer trigger the salivation neuron; on the other hand, when the salivation neuron is stimulated by the ring neuron alone, at first the salivation neuron can be triggered but after the salivation neuron realizes that the “ringing” is not associated with “food,” the salivation neuron will not be triggered any longer. How to integrate the proposed network into large scale memristive neural network with multiple associations is also introduced. Simulations results demonstrate the correctness of the designs.

Energy‐Efficient Hybrid Perovskite Memristors and Synaptic Devices

New parallel computing architectures based on neuromorphic computing are needed due to their advantages over conventional computation with regards to real‐time processing of unstructured sensory data such as image, video, or voice. However, developing artificial neuromorphic system remains a challenge due to the lack of electronic synaptic devices, which can mimic all the functions of biological synapses with low energy consumption. Here it is reported that two‐terminal organometal trihalide perovskite (OTP) synaptic devices can mimic the neuromorphic learning and remembering process. Various functions known in biological synapses are demonstrated in OTP synaptic devices including four forms of spike‐timing‐dependent plasticity (STDP), spike‐rate‐dependent plasticity (SRDP), short‐term plasticity (STP) and long‐term potentiation (LTP)), and learning‐experience behavior. The excellent photovoltaic property of the OTP devices also enables photo‐read synaptic functions. The perovskite synapse has the potential of low energy consumption of femto‐Joule/(100 nm)2 per event, which is close to the energy consumption of biological synapses. The demonstration of energy‐efficient OTP synaptic devices opens a new plausible application of OTP materials into neuromorphic devices, which offer the high connectivity and high density required for biomimic computing.

An improved WOx memristor model with synapse characteristic analysis

Memristor is defined as the fourth basic electronic element, the studies on its models exhibit diversity. Now, the matching extent between memristor model and natural memristor has received researchers' wide attention. A new memristor model is proposed by changing the ion diffusion term of the WOx memristor, namely, adding another two internal state variables τ and μ which denote the relaxation time and retention, respectively, and the improved model can simulate natural memristor better. Firstly, the new one is able to not only describe the general characteristics of a memrsitor, but also capture the memory loss behavior. In addition, the new memristor can be considered as a neural synapse, under the action of the input pulses with different amplitudes, duration and intervals, the spike rate dependent plasticity, short-term plasticity (STP), and long-term plasticity (LTP) are analyzed, and the ''learning experience'' phenomenon which is very similar to the biological system is discovered, most of which is due to the back diffusion of the oxygen vacancies during the intervals of the input pulses which are caused by the concentration difference. Moreover, an exponential decay equation is built to describe the relaxation process of STP. Finally, taking into consideration the relationship between temperature and ion diffusion coefficient, the effect of temperature on the relaxation process of STP is discussed. Experimental results show that the new memristor model can better match the actual behavior characteristics, and more suitably acts as a synapse for being applied to neuromorphic systems.

Polymer memristor for information storage and neuromorphic applications

Polymermaterials have been considered as promising candidates for the implementation of memristor devices due to their low-cost, easy solution processability, mechanical flexibility and ductibility, tunable electronic performance through innovative molecular design cum synthesis strategy and compatibility with complementary metal oxide semiconductor (CMOS) technology as well. The digital-type polymer memristor behaves as resistive random access memory with non-volatility, high density, more speed, low power consumption, large ON/OFF ratio, high endurance and long retention, and is recognized as an appealing candidate for the next generation universal memory. As a logic component, the analog-type memristor, with the ability to emulate the fundamental synaptic functions of short-term/long-term plasticity (STP/LTP), spike-timing dependent-plasticity (STDP), spike-rate dependent plasticity (SRDP) and learningexperience behaviors, can be used to construct artificial neural networks for neuromorphic computation. In this review, we shall attempt to summarize the recent progress in research on the materials, switching characteristics and mechanism aspects of two terminal polymer memristors, for both information storage and neuromorphic applications that inspire great interest in the industrial and academic communities.

Enabling an integrated rate-temporal learning scheme on memristor.

Learning scheme is the key to the utilization of spike-based computation and the emulation of neural/synaptic behaviors toward realization of cognition. The biological observations reveal an integrated spike time- and spike rate-dependent plasticity as a function of presynaptic firing frequency. However, this integrated rate-temporal learning scheme has not been realized on any nano devices. In this paper, such scheme is successfully demonstrated on a memristor. Great robustness against the spiking rate fluctuation is achieved by waveform engineering with the aid of good analog properties exhibited by the iron oxide-based memristor. The spike-time-dependence plasticity (STDP) occurs at moderate presynaptic firing frequencies and spike-rate-dependence plasticity (SRDP) dominates other regions. This demonstration provides a novel approach in neural coding implementation, which facilitates the development of bio-inspired computing systems.

A biophysically-based neuromorphic model of spike rate- and timing-dependent plasticity

Current advances in neuromorphic engineering have made it possible to emulate complex neuronal ion channel and intracellular ionic dynamics in real time using highly compact and power-efficient complementary metal-oxide-semiconductor (CMOS) analog very-large-scale-integrated circuit technology. Recently, there has been growing interest in the neuromorphic emulation of the spike-timing-dependent plasticity (STDP) Hebbian learning rule by phenomenological modeling using CMOS, memristor or other analog devices. Here, we propose a CMOS circuit implementation of a biophysically grounded neuromorphic (iono-neuromorphic) model of synaptic plasticity that is capable of capturing both the spike rate-dependent plasticity (SRDP, of the Bienenstock-Cooper-Munro or BCM type) and STDP rules. The iono-neuromorphic model reproduces bidirectional synaptic changes with NMDA receptor-dependent and intracellular calcium-mediated long-term potentiation or long-term depression assuming retrograde endocannabinoid signaling as a second coincidence detector. Changes in excitatory or inhibitory synaptic weights are registered and stored in a nonvolatile and compact digital format analogous to the discrete insertion and removal of AMPA or GABA receptor channels. The versatile Hebbian synapse device is applicable to a variety of neuroprosthesis, brain-machine interface, neurorobotics, neuromimetic computation, machine learning, and neural-inspired adaptive control problems.