Spike-timing-dependent Plasticity Research Articles

Interaction between Theta Phase and Spike Timing-Dependent Plasticity Simulates Theta-Induced Memory Effects.

Rodent studies suggest that spike timing relative to hippocampal theta activity determines whether potentiation or depression of synapses arise. Such changes also depend on spike timing between presynaptic and postsynaptic neurons, known as spike timing-dependent plasticity (STDP). STDP, together with theta phase-dependent learning, has inspired several computational models of learning and memory. However, evidence to elucidate how these mechanisms directly link to human episodic memory is lacking. In a computational model, we modulate long-term potentiation (LTP) and long-term depression (LTD) of STDP, by opposing phases of a simulated theta rhythm. We fit parameters to a hippocampal cell culture study in which LTP and LTD were observed to occur in opposing phases of a theta rhythm. Further, we modulated two inputs by cosine waves with 0° and asynchronous phase offsets and replicate key findings in human episodic memory. Learning advantage was found for the in-phase condition, compared with the out-of-phase conditions, and was specific to theta-modulated inputs. Importantly, simulations with and without each mechanism suggest that both STDP and theta phase-dependent plasticity are necessary to replicate the findings. Together, the results indicate a role for circuit-level mechanisms, which bridge the gap between slice preparation studies and human memory.

Ultra-low operation current and abnormal bipolar switching phenomena of hydrogen-passivated HfO2 memristive devices for low power artificial neural network applications

We successfully demonstrated a hydrogen (H2)-passivated HfO2-based memristive device that operates with a low current that is below sub-nA in this paper. Gradual resistance changes are achieved with the diversification of the sweeping voltage. The H2-passivated HfO2-based memristive device shows an improvement of the device's performance of low power consumption with sub-nA, data retention for 104 s, and robust endurance of 106 repetitive cycles. The Spike Timing Dependent Plasticity (STDP) and the pattern recognition via MNIST are employed to assess the artificial neural network characteristics of the proposed device. A symmetrical curve on STDP is observed for an H2-passivated HfO2-based memristive device, which indicates that it can facilitate in order to mimic a human brain. The high accuracy of 91.12% and a clearer pattern recognition are obtained on an H2-passivated memristive device. Therefore, we believe that the H2-passivated HfO2-based memristive device opens up a new path in order to mimic the human brain with low power.

Transcranial cortico-cortical paired associative stimulation (ccPAS) over ventral premotor-motor pathways enhances action performance and corticomotor excitability in young adults more than in elderly adults.

Transcranial magnetic stimulation (TMS) methods such as cortico-cortical paired associative stimulation (ccPAS) can increase the strength of functional connectivity between ventral premotor cortex (PMv) and primary motor cortex (M1) via spike timing-dependent plasticity (STDP), leading to enhanced motor functions in young adults. However, whether this STDP-inducing protocol is effective in the aging brain remains unclear. In two groups of young and elderly healthy adults, we evaluated manual dexterity with the 9-hole peg task before and after ccPAS of the left PMv-M1 circuit. We observed that ccPAS enhanced dexterity in young adults, and this effect was anticipated by a progressive increase in motor-evoked potentials (MEPs) during ccPAS administration. No similar effects were observed in elderly individuals or in a control task. Across age groups, we observed that the magnitude of MEP changes predicted larger behavioral improvements. These findings demonstrate that left PMv-to-M1 ccPAS induces functionally specific improvements in young adults' manual dexterity and an increase in corticomotor excitability, but altered plasticity prevents the effectiveness of ccPAS in the elderly.

A shallow hybrid classical–quantum spiking feedforward neural network for noise-robust image classification

Deep Convolutional Neural Network (CNN)-based image classification systems are often susceptible to noise interruption, i.e., minor image noise may significantly impact the outcome. On the contrary, classical Spiking Neural Network (SNN) is known for handling noisy data due to the stochastic and temporal behaviour of the spiking neuron signals. However, it is not always feasible to train the weights of the classical SNN due to stochastic and non-differentiable spiking signals. Recent applications of quantum machine learning to stochastic modelling have predicted that quantum computing would prove to be a significant advantage, hence catapulting the study of quantum and neuromorphic computing to a new level of development. Motivated by these observations, this paper introduces a shallow hybrid classical–quantum spiking feedforward neural network referred to as Spiking Quantum Neural Network (SQNN) for dealing with a robust image classification task in the presence of noise and adversarial attacks. The proposed SQNN offers the inherent capabilities of processing unforeseen noisy test images due to its spatial and temporal information. An efficient variational quantum circuit training with the help of a standard back-propagation algorithm obviates the classical Spiking Time-Dependent Plasticity (STDP) and Spike-Prop algorithms, which are often found inefficient in training the feedforward SNN. The proposed SQNN is extensively tested and benchmarked on the PennyLane Quantum Simulator. Experimental results show that the proposed SQNN model supersedes the feedforward SNN (SFNN), Random Quantum Neural Networks (RQNN), AlexNet and ResNet-18 on unseen test images with added noise from the FashionMNIST, MNIST, KMNIST, CIFAR10, and ImageNet datasets.

Decoupling of interacting neuronal populations by time-shifted stimulation through spike-timing-dependent plasticity.

The synaptic organization of the brain is constantly modified by activity-dependent synaptic plasticity. In several neurological disorders, abnormal neuronal activity and pathological synaptic connectivity may significantly impair normal brain function. Reorganization of neuronal circuits by therapeutic stimulation has the potential to restore normal brain dynamics. Increasing evidence suggests that the temporal stimulation pattern crucially determines the long-lasting therapeutic effects of stimulation. Here, we tested whether a specific pattern of brain stimulation can enable the suppression of pathologically strong inter-population synaptic connectivity through spike-timing-dependent plasticity (STDP). More specifically, we tested how introducing a time shift between stimuli delivered to two interacting populations of neurons can effectively decouple them. To that end, we first used a tractable model, i.e., two bidirectionally coupled leaky integrate-and-fire (LIF) neurons, to theoretically analyze the optimal range of stimulation frequency and time shift for decoupling. We then extended our results to two reciprocally connected neuronal populations (modules) where inter-population delayed connections were modified by STDP. As predicted by the theoretical results, appropriately time-shifted stimulation causes a decoupling of the two-module system through STDP, i.e., by unlearning pathologically strong synaptic interactions between the two populations. Based on the overall topology of the connections, the decoupling of the two modules, in turn, causes a desynchronization of the populations that outlasts the cessation of stimulation. Decoupling effects of the time-shifted stimulation can be realized by time-shifted burst stimulation as well as time-shifted continuous simulation. Our results provide insight into the further optimization of a variety of multichannel stimulation protocols aiming at a therapeutic reshaping of diseased brain networks.

Modulating STDP With Back-Propagated Error Signals to Train SNNs for Audio Classification

Audio classification has many practical applications such as noise pollution detection, wildlife monitoring, speech recognition, and more. For many of these applications, deploying classifiers on low powered devices for persistent deployment is desirable. Artificial neural networks (ANN) have achieved state-of-the-art performance on audio classification tasks; however, it is not always feasible to deploy modern ANNs to embedded devices due to their high power consumption. Biologically inspired spiking neural networks (SNN) have been shown to significantly reduce power consumption during inference when compared with equivalent ANNs, and they have also been theoretically proven to be more computationally powerful than stateless ANNs. This work proposes an audio classification system using SNNs, and a learning algorithm is developed for classification with multilayer SNNs which combines biologically plausible spike-timing-dependent plasticity (STDP) with spatial error backpropagation. By allowing the STDP process to account for both temporal dependencies and the non-differentiable activation function derivative, the proposed learning rule successfully trains multilayer SNNs for the considered classification tasks. Through the STDP process, the proposed learning rule is also capable of online learning; explicit storage of values from previous timesteps is not required, in contrast to the widely adopted backpropagation through time (BPTT) algorithm. The proposed approach approaches the performance of SNNs trained via BPTT on the classification tasks. SNNs trained with the proposed learning rule are evaluated on the Iris Flower dataset, the Real-World Computing Partnership sounds dataset, the Free Spoken Digits Dataset, and the UrbanSound8k dataset.

Flexible Memristor Constructed by 2D Cadmium Phosphorus Trichalcogenide for Artificial Synapse and Logic Operation (Adv. Funct. Mater. 9/2023)

Transition-Metal Phosphorus Trichalcogenide In article number 2211269, Jiahong Wang, Cong Ye, and co-workers report a flexible memristor fabricated utilizing two-dimensional cadmium phosphorus trichalcogenide nanosheets; the synaptic plasticity, including paired-pulse facilitation and spiking timing-dependent plasticity, are further observed. Owing to the linear conductance modulation capacity, the decimal operation is explored.

An Area- and Energy-Efficient Spiking Neural Network With Spike-Time-Dependent Plasticity Realized With SRAM Processing-in-Memory Macro and On-Chip Unsupervised Learning.

In this article, we present a spiking neural network (SNN) based on both SRAM processing-in-memory (PIM) macro and on-chip unsupervised learning with Spike-Time-Dependent Plasticity (STDP). Co-design of algorithm and hardware for hardware-friendly SNN and efficient STDP-based learning methodology is used to improve area and energy efficiency. The proposed macro utilizes charge sharing of capacitors to perform fully parallel Reconfigurable Multi-bit PIM Multiply-Accumulate (RMPMA) operations. A thermometer-coded Programmable High-precision PIM Threshold Generator (PHPTG) is designed to achieve low differential non-linearity (DNL) and high linearity. In the macro, each column of PIM cells and a comparator act as a neuron to accumulate membrane potential and fire spikes. A simplified Winner Takes All (WTA) mechanism is used in the proposed hardware-friendly architecture. By combining the hardware-friendly STDP algorithm as well as the parallel Word Lines (WLs) and Processing Bit Lines (PBLs), we realize unsupervised learning and recognize the Modified National Institute of Standards and Technology (MNIST) dataset. The chip for the hardware implementation was fabricated with a 55 nm CMOS process. The measurement shows that the chip achieves a learning efficiency of 0.47 nJ/pixel, with a learning energy efficiency of 70.38 TOPS/W. This work paves a pathway for the on-chip learning algorithm in PIM with lower power consumption and fewer hardware resources.

Reconfigurable logic gates in biological crossbar neural networks using STDP learning.

McCulloch and Pitts provided a way to describe brain functions in logic gates. The idea of Boolean neurons made a tremendous contribution to the development of artificial intelligence. However, logic gates based on biological neural networks have never been implemented. To investigate where logic gates can be demonstrated using biological neural networks. We constructed biological crossbar neural networks on multi-electrodes arrays (MEA) using PDMS microfluidic channels. The fluorescence optical images revealed that crossbar array was connected by axons. We measured the activation curves using the stimulus-evoked spiking probability, and found that the threshold voltage of activation curves were affected by the spike-timing-dependent-plasticity (STDP) learning. As a result, reconfigurable logic gates of AND and OR could be successfully implemented through STDP learning. Many neuromorphic systems have the crossbar array structure; thus, our study may contribute to the development of neuromorphic research.

Heterogeneous recurrent spiking neural network for spatio-temporal classification.

Spiking Neural Networks are often touted as brain-inspired learning models for the third wave of Artificial Intelligence. Although recent SNNs trained with supervised backpropagation show classification accuracy comparable to deep networks, the performance of unsupervised learning-based SNNs remains much lower. This paper presents a heterogeneous recurrent spiking neural network (HRSNN) with unsupervised learning for spatio-temporal classification of video activity recognition tasks on RGB (KTH, UCF11, UCF101) and event-based datasets (DVS128 Gesture). We observed an accuracy of 94.32% for the KTH dataset, 79.58% and 77.53% for the UCF11 and UCF101 datasets, respectively, and an accuracy of 96.54% on the event-based DVS Gesture dataset using the novel unsupervised HRSNN model. The key novelty of the HRSNN is that the recurrent layer in HRSNN consists of heterogeneous neurons with varying firing/relaxation dynamics, and they are trained via heterogeneous spike-time-dependent-plasticity (STDP) with varying learning dynamics for each synapse. We show that this novel combination of heterogeneity in architecture and learning method outperforms current homogeneous spiking neural networks. We further show that HRSNN can achieve similar performance to state-of-the-art backpropagation trained supervised SNN, but with less computation (fewer neurons and sparse connection) and less training data.

An Efficient Design of TaOx‐Based Memristor by Inserting an Ultrathin Al2O3 Layer with High Stability for Neuromorphic Computing and Logic Operation

AbstractNew computing‐in‐memory architecture based on memristors can achieve in situ storage and computing of data, which greatly improves the computing efficiency of the hardware system. Here, a reliable bilayer structured TaOx/Al2O3 memristor with a 2 nm Al2O3 insertion layer is demonstrated. This device exhibits stable and gradual switching behavior with a low set/reset voltage (0.61 V/−0.49 V) and multilevel conductance characteristics. It is further indicated that the device has a larger ON/Off ratio (≈148×) and better nonlinearity of conductance modulation by inserting an Al2O3 layer. Various forms of synaptic plasticity are mimicked, such as long‐term potentiation/depression (LTP/LTD), paired‐pulse facilitation (PPF), and spike‐timing‐dependent plasticity (STDP). Based on the quasi‐linear conductance modulation characteristics, excellent classification accuracy (90.4%) is achieved for the applications of handwritten digit recognition. Moreover, the logic operations (intersection, union, and complement) are implemented on a 3 × 5 memristor array, which shows an efficient way to design versatile and reliable devices and provides a novel idea for neuromorphic computing and in‐memory logic operation.

All oxide based flexible multi-folded invisible synapse as vision photo-receptor

All oxide-based transparent flexible memristor is prioritized for the potential application in artificially simulated biological optoelectronic synaptic devices. SnOx memristor with HfOx layer is found to enable a significant effect on synaptic properties. The memristor exhibits good reliability with long retention, 104 s, and high endurance, 104 cycles. The optimized 6 nm thick HfOx layer in SnOx-based memristor possesses the excellent synaptic properties of stable 350 epochs training, multi-level conductance (MLC) behaviour, and the nonlinearity of 1.53 and 1.46 for long-term potentiation and depression, respectively, and faster image recognition accuracy of 100% after 23 iterations. The maximum weight changes of -73.12 and 79.91% for the potentiation and depression of the synaptic device, respectively, are observed from the spike-timing-dependent plasticity (STDP) characteristics making it suitable for biological applications. The flexibility of the device on the PEN substrate is confirmed by the acceptable change of nonlinearities up to 4 mm bending. Such a synaptic device is expected to be used as a vision photo-receptor.

Interface engineering of amorphous gallium oxide crossbar array memristors for neuromorphic computing

This paper reports on the fabrication and characterization of crossbar array memristors using amorphous gallium oxide (a-GaO x ) for implementing high-speed and wide-dynamic range artificial synaptic functions. The a-GaO x memristors were fabricated by pulsed laser deposition in an argon atmosphere using a platinum bottom electrode and an indium tin oxide (ITO) top electrode. We revealed that the interface engineering at a-GaO x /ITO is the key to demonstrating exemplary resistive switching operation. Stable counter figure-8 hysteresis loops were obtained by voltage application, leading to the successful demonstration of non-volatile retention over 104 s and the multi-level conductance modulation. Furthermore, spike-timing-dependent plasticity (STDP) was artificially implemented by applying pre- and post-spike voltages to the device. Consequently, significant weight-change rates were achieved in the asymmetric STDP imitation, which can be attributed to the reliable resistive switching properties of the device with an extensive dynamic range. These results indicate that the a-GaO x crossbar array memristor is a promising hardware platform for neuromorphic computing applications.

Synaptic and Gradual Conductance Switching Behaviors in CeO2/Nb–SrTiO3 Heterojunction Memristors for Electrocardiogram Signal Recognition

The synaptic properties of memristors have been widely studied. However, researchers are still committed to solving various challenges, including the study of highly reliable memristors with comprehensive synaptic functions and memristors that simulate highly complex neurological learning rules. In this work, we report a CeO2/Nb-SrTiO3 heterojunction memristor whose conductance could be gradually tuned under both positive and negative pulse trains. Due to the gradual conductance switching behavior and the high switching ratio (105), the CeO2/Nb-SrTiO3 heterojunction memristor could dutifully mimic biosynaptic functions, including excitatory/inhibitory postsynaptic current (EPSC/IPSC), paired-pulse facilitation and depression (PPF/PPD), spike amplitude-dependent plasticity (SADP), spike duration-dependent plasticity (SDDP), spike rate-dependent plasticity (SRDP), paired/triplet spiking-time-dependent plasticity (STDP), and Bienenstock-Cooper-Munro (BCM) rules. Moreover, a convolutional neural network based on the memristors is constructed to identify the electrocardiogram (ECG) data sets to realize the diagnosis of diseases with a recognition accuracy of 93%. Besides, the recognition accuracy of the handwriting digit reaches 96%. These studies broaden the research scope of high-level synaptic behavior and lay a foundation for the future full synaptic memristor networks.

Titanium oxide memristors driven by oxygen diffusion dynamics and 1S1M biomimetic system

Temperature-sensitive properties of artificial electronic synapses allow the devices to have both bionanomimetic and sensor functions, which are promising for the research of improving bio-neuro-bionic systems. In this article, an ITO/TiOy/TiOx/TiN memristor with a stacked layer structure was prepared to achieve a gradual modulation of the operating current by depositing the oxygen-deficient layer TiOy. The devices exhibit clear sensitivity and stable retention over the temperature range of 299K–349K for 1000 s, while the 1000 endurance cycles allow for biosynaptic analog modulation capabilities, including paired-pulse facilitation (PPF) and spike-timing-dependent plasticity (STDP) at room temperature. The mechanism of conductivity based on oxygen diffusion dynamics can be well demonstrated by high-resolution transmission electron microscopy (HRTEM) and X-ray photoelectron spectroscopy (XPS) analysis. Furthermore, the sensing-storage integrated 1S1M system based on the ITO/TiOy/TiOx/TiN memristor provides a reasonable conception for the bionic research of biological nervous systems.

Paired associative stimulation on the soleus H-Reflex using motor point and peripheral nerve stimulation

Paired associative stimulation (PAS) has been shown to modulate the corticospinal excitability via spike timing dependent plasticity (STDP). In this study, we aimed to suppress the spinal H-Reflex using PAS. We paired two stimulation modalities, i.e., peripheral nerve stimulation (PNS) and motor point stimulation (MPS). We used PNS to dominantly activate the Ia sensory axon, and we used MPS to dominantly activate the α-motoneuron cell body antidromically. Thus, we applied both PNS and MPS such that the α-motoneuron cell body was activated 5 ms before the activation of the Ia sensory axon ending at the Ia-α motoneuron synapse. If the spinal reflexes can be modulated by STDP, and a combination of MPS and PNS is timed appropriately, then the H-Reflex amplitude will decrease while no change in H-Reflex amplitude is expected for MPS or PNS only. To test this hypothesis, six young healthy participants (5M/1F: 26.8 ± 4.1 yrs) received one of the three following conditions on days separated by at least 24 hr: 1) PAS, 2) MPS only or 3) PNS only. The H-Reflex and M-wave recruitment curves of the soleus were measured immediately prior to, immediately after, 30 min and 60 min after the intervention. The normalized H-Reflex amplitudes were then compared across conditions and times using a two-way ANOVA (3 conditions × 4 times). No main effects of condition or time, or interaction effect were found. These results suggest that relying solely on STDP may be insufficient to inhibit the soleus H-Reflex.

Coherence resonance and stochastic synchronization in a small-world neural network: an interplay in the presence of spike-timing-dependent plasticity

Coherence resonance (CR), stochastic synchronization (SS), and spike-timing-dependent plasticity (STDP) are ubiquitous dynamical processes in biological neural networks. Whether there exists an optimal network and STDP configuration at which CR and SS are both pronounced is a fundamental question of interest that is still elusive. We expect such a configuration to enable the brain to make synergistic and optimal use of these phenomena to process information efficiently. This paper considers a small-world network of excitable Hodgkin–Huxley neurons driven by channel noise and STDP with an asymmetric Hebbian time window. Numerical results indicate specific network topology and STDP parameter intervals in which CR and SS can be simultaneously enhanced. Our results imply that an optimally tuned inherent background noise, STDP rule, and network topology can play a constructive role in enhancing both the time precision of firing and the synchronization in neural systems.

Diverse synaptic weight adjustment of bio-inspired ZrOx-based memristors for neuromorphic system

In this article, we demonstrate the bio-inspired synaptic features of the TiN/ZrOx/Pt capacitor structure for neuromorphic engineering. The chemical and material compositions and the thicknesses of each of the layers are verified by using transmission electron microscopy (TEM) images and energy-dispersive X-ray spectroscopy (EDS) maps. Stable resistive switching with a low set voltage (∼1 V) was determined by scanning the DC I–V curves of many cells. The DC endurance of ∼104 cycles and retention (10,000 s) in five states was achieved. Multi-level cells (MLC) characteristics were achieved based on the compliance current and reset stop voltage in DC sweep and pulses. Finally, we emulated paired-pulse facilitation (PPF), paired-pulse depression (PPD), electric excitatory postsynaptic current (EPSC), and spike-timing-dependent plasticity (STDP) of the artificial synapse by using the RRAM device.

Unsupervised and efficient learning in sparsely activated convolutional spiking neural networks enabled by voltage-dependent synaptic plasticity

Spiking neural networks (SNNs) are gaining attention due to their energy-efficient computing ability, making them relevant for implementation on low-power neuromorphic hardware. Their biological plausibility has permitted them to benefit from unsupervised learning with bio-inspired plasticity rules, such as spike timing-dependent plasticity (STDP). However, standard STDP has some limitations that make it challenging to implement on hardware. In this paper, we propose a convolutional SNN (CSNN) integrating single-spike integrate-and-fire (SSIF) neurons and trained for the first time with voltage-dependent synaptic plasticity (VDSP), a novel unsupervised and local plasticity rule developed for the implementation of STDP on memristive-based neuromorphic hardware. We evaluated the CSNN on the TIDIGITS dataset, where, helped by our sound preprocessing pipeline, we obtained a performance better than the state of the art, with a mean accuracy of 99.43%. Moreover, the use of SSIF neurons, coupled with time-to-first-spike (TTFS) encoding, results in a sparsely activated model, as we recorded a mean of 5036 spikes per input over the 172 580 neurons of the network. This makes the proposed CSNN promising for the development of models that are extremely efficient in energy. We also demonstrate the efficiency of VDSP on the MNIST dataset, where we obtained results comparable to the state of the art, with an accuracy of 98.56%. Our adaptation of VDSP for SSIF neurons introduces a depression factor that has been very effective at reducing the number of training samples needed, and hence, training time, by a factor of two and more, with similar performance.

Spatial Computing in Modular Spiking Neural Networks with a Robotic Embodiment

One of the challenges in modern neuroscience is creating a brain-on-a-chip. Such a semiartificial device based on neural networks grown in vitro should interact with the environment when embodied in a robot. A crucial point in this endeavor is developing a neural network architecture capable of associative learning. This work proposes a mathematical model of a midscale modular spiking neural network (SNN) to study learning mechanisms within the brain-on-a-chip context. We show that besides spike-timing-dependent plasticity (STDP), synaptic and neuronal competitions are critical factors for successful learning. Moreover, the shortest pathway rule can implement the synaptic competition responsible for processing conditional stimuli coming from the environment. This solution is ready for testing in neuronal cultures. The neuronal competition can be implemented by lateral inhibition actuating over the SNN modulus responsible for unconditional responses. Empirical testing of this approach is challenging and requires the development of a technique for growing cultures with a given ratio of excitatory and inhibitory neurons. We test the modular SNN embedded in a mobile robot and show that it can establish the association between touch (unconditional) and ultrasonic (conditional) sensors. Then, the robot can avoid obstacles without hitting them, relying on ultrasonic sensors only.