Plasticity Learning Rule Research Articles

Mimicking Neurotransmitter Activity and Realizing Algebraic Arithmetic on Flexible Protein-Gated Oxide Neuromorphic Transistors.

Recently, flexible neuromorphic devices have attracted extensive attention for the construction of perception cognitive systems with the ultimate objective to achieve robust computation, efficient learning, and adaptability to evolutionary changes. In particular, the design of flexible neuromorphic devices with data processing and arithmetic capabilities is highly desirable for wearable cognitive platforms. Here, an albumen-based protein-gated flexible indium tin oxide (ITO) ionotronic neuromorphic transistor was proposed. First, the transistor demonstrates excellent mechanical robustness against bending stress. Moreover, spike-duration-dependent synaptic plasticity and spike-amplitude-dependent synaptic plasticity behaviors are not affected by bending stress. With the unique protonic gating behaviors, neurotransmission processes in biological synapses are emulated, exhibiting three characteristics in neurotransmitter release, including quantal release, stochastic release, and excitatory or inhibitory release. In addition, three types of spike-timing-dependent plasticity learning rules are mimicked on the ITO ionotronic neuromorphic transistor. Most interestingly, algebraic arithmetic operations, including addition, subtraction, multiplication, and division, are implemented on the protein gated neuromorphic transistor for the first time. The present work would open a promising biorealistic avenue to the scientific community to control and design wearable "green" cognitive platforms, with potential applications including but not limited to intelligent humanoid robots and replacement neuroprosthetics.

STDP Based Unsupervised Multimodal Learning With Cross-Modal Processing in Spiking Neural Networks

Spiking neural networks perform reasonably well in recognition applications for single modality (e.g., images, audio, or text). In this paper, we propose a multimodal spiking neural network that combines two modalities (image and audio). The two unimodal ensembles are connected with cross-modal connections and the entire network is trained with unsupervised learning. The network receives inputs in both modalities for the same class and predicts the class label. The excitatory connections in the unimodal ensemble and the cross-modal connections are trained with power-law weight-dependent spike timing dependent plasticity learning rule. The cross-modal connections capture the correlation between neurons of different modalities. The multimodal network learns features of both modalities and improves the classification accuracy compared to unimodal topology, even when one of the modality is distorted by noise. The cross-modal connections suppress the effect of noise on classification accuracy. The well-learned cross-modal connections invoke additional spiking activity in neurons of the correct label. The cross-modal connections are only excitatory and do not inhibit the normal activity of the unimodal ensembles. We evaluated our multimodal network on images from MNIST dataset and utterances of digits from TI46 speech corpus. The multimodal network achieved a classification accuracy of 98% on the combined MNIST and TI46 dataset.

Gene essentiality prediction based on chaos game representation and spiking neural networks

Abstract Chaos game representation (CGR) is a useful one-to-one visualization tool to represent nucleotide sequences, in which both local and global patterns of nucleotides can be graphically described. Deep learning networks have been proved to achieve outstanding performance on feature extraction and image recognition. In this paper, we use convolutional spiking neural networks (SNNs) with reward-modulated spike-timing-dependent plasticity (R-STDP) learning rule to learn from the frequency matrix chaos game representation (FCGR) images of essential and non-essential genes of 32 bacteria in the DEG database and make intra-organism and cross-organism essential gene predictions. For intra-organism predictions, our highest accuracy(ACC) score is 0.90 and the average ACC is 0.78, and for cross-organism predictions, our highest ACC is 0.79 and the average ACC is 0.68. Compared with the results of traditional machine learning classifiers training with FCGR images or numerical fractal features pre-calculated from CGR representations, our intra-organism prediction results are much better for all the bacteria or most bacteria, respectively, indicating that our spiking neural networks can make better essential gene prediction by extracting the gene features directly from the FCGR images of essential and nonessential genes. Compared with essential gene prediction methods using gene sequence features and topological features, our cross-organism prediction results can achieve performance close to or even better than such methods, while requiring much fewer input features.

Synergies between synaptic and intrinsic plasticity in echo state networks

Synaptic plasticity and intrinsic plasticity, as two of the most common neural plasticity mechanisms, occur in all neural circuits throughout life. Neurobiological studies indicated that the interplay between synaptic and intrinsic plasticity contributes to the adaptation of the nervous system to different synaptic input signals. However, most existing computational models of neural plasticity consider these two plasticity mechanisms separately, which is biologically implausible. In this paper, a synergistic plasticity learning rule is proposed to adapt the reservoir connections in echo state networks (ESNs), which not only takes into account the regulation of synaptic weights, but also considers the adjustment of neuronal intrinsic excitability. The proposed synergetic plasticity rule is verified on a number of prediction and classification benchmark problems and our empirical results demonstrate that the ESN with synergistic plasticity learning rule performs much better than the state-of-the-art ESN models, and an ESN with a single neural plasticity rule.

Synaptic plasticity and classical conditioning mimicked in single indium-tungsten-oxide based neuromorphic transistor* *Project supported by the National Natural Science Foundation of China (Grant Nos. 11674162 and 61834001) and the National Key R&D Program of China (Grant Nos. 2018YFA0305800 and 2019YFB2205400).

Emulation of synaptic function by ionic/electronic hybrid device is crucial for brain-like computing and neuromorphic systems. Electric-double-layer (EDL) transistors with proton conducting electrolytes as the gate dielectrics provide a prospective approach for such application. Here, artificial synapses based on indium-tungsten-oxide (IWO)-based EDL transistors are proposed, and some important synaptic functions (excitatory post-synaptic current, paired-pulse facilitation, filtering) are emulated. Two types of spike-timing-dependent plasticity (Hebbian STDP and anti-Hebbian STDP) learning rules and multistore memory (sensory memory, short-term memory, and long-term memory) are also mimicked. At last, classical conditioning is successfully demonstrated. Our results indicate that IWO-based neuromorphic transistors are interesting for neuromorphic applications.

An ensemble unsupervised spiking neural network for objective recognition

It is now known that the spiking neuron is a basic unit of spiking neural networks (SNNs). Spiking neurons modulate the nervous cells via receiving external incentives, generation of action potential and firing spikes. The SNNs usually used for pattern recognition tasks or complex computation depending on the brain-like characteristic. Although the SNNs have no advantages comparing with the deep neural networks in terms of classification accuracy, the SNNs have more characteristics of biological neurons. In this paper, a hierarchical SNN, comprising convolutional and pooling layers, is designed. The proposed SNN consists of excitatory and inhibitory neurons based on the mechanism of the primate brain. A temporal coding (rank order) manner is used to encode the input patterns. It depends on the rank of the spike arrival on post synapses to establish the priority of input spikes for a particular pattern. The spike-timing-dependent plasticity (STDP) learning rule is used in convolutional layers to extract visual features in an unsupervised learning manner. During the classification stage, a lateral inhibition mechanism is used to prevent the non-firing neurons and produce distinguishable results. In order to improve the performance of our SNN, an ensemble SNN architecture using the voting method is proposed, and transfer learning is used to avoid re-training the SNN when solving the different tasks. The hand-written digits classification task on MNIST, CIFAR-10, and BreaKHis databases are used to verify the performance of the proposed SNN. Experimental results show that by using the ensemble architecture and transfer learning, the classification accuracy of 99.27% for the MNIST database, overall accuracy is 93% for the CIFAR-10 database, and overall accuracy is 96.97% for BreaKHis database. In the meantime, this work achieves a better performance than the benchmarking approaches. Taken together, the results of our work suggest that the ensemble SNN architecture with transfer learning is key to improving the performance of the SNN.

Temporal-Sequential Learning With a Brain-Inspired Spiking Neural Network and Its Application to Musical Memory.

Sequence learning is a fundamental cognitive function of the brain. However, the ways in which sequential information is represented and memorized are not dealt with satisfactorily by existing models. To overcome this deficiency, this paper introduces a spiking neural network based on psychological and neurobiological findings at multiple scales. Compared with existing methods, our model has four novel features: (1) It contains several collaborative subnetworks similar to those in brain regions with different cognitive functions. The individual building blocks of the simulated areas are neural functional minicolumns composed of biologically plausible neurons. Both excitatory and inhibitory connections between neurons are modulated dynamically using a spike-timing-dependent plasticity learning rule. (2) Inspired by the mechanisms of the brain's cortical-striatal loop, a dependent timing module is constructed to encode temporal information, which is essential in sequence learning but has not been processed well by traditional algorithms. (3) Goal-based and episodic retrievals can be achieved at different time scales. (4) Musical memory is used as an application to validate the model. Experiments show that the model can store a huge amount of data on melodies and recall them with high accuracy. In addition, it can remember the entirety of a melody given only an episode or the melody played at different paces.

Presynaptic Spike-Driven Spike Timing-Dependent Plasticity With Address Event Representation for Large-Scale Neuromorphic Systems

Learning plays an important role in the brain to make it adaptive to dynamical environments. This paper presents a presynaptic spike-driven spike timing-dependent plasticity (STDP) learning rule in the address domain for a neuromorphic architecture using a synaptic connectivity table in an external memory at a local routing node. We contribute two aspects to the implementation of the learning rule for extended large-scale neuromorphic systems. First, we reduced buffer sizes required for tracing a spike train which is required to pair all presynaptic and postsynaptic spike for an STDP time window. This method implements an exponential decay STDP function with two parameters: the latest timestamp and the synaptic modification rate at the latest timestamp. It reduces the required buffer size compared to previous works. Second, we resolve a lack of reverse lookup table issue with the presynaptic spike-driven algorithm. The proposed algorithm holds causal updates at postsynaptic spikes until a next presynaptic spike arrival. This approach removes the need of a reverse lookup table required at a postsynaptic spike. We show the implementation of the proposed algorithm in an FPGA device and validate it with a spiking neural network configuration. The experiment results show the proposed algorithm is comparable qualitatively with a conventional STDP learning rule.

Digital multiplier‐less implementation of high‐precision SDSP and synaptic strength‐based STDP

SummarySpiking neural networks (SNNs) can achieve lower latency and higher efficiency compared with traditional neural networks if they are implemented in dedicated neuromorphic hardware. In both biological and artificial spiking neuronal systems, synaptic modifications are the main mechanism for learning. Plastic synapses are thus the core component of neuromorphic hardware with on‐chip learning capability. Recently, several research groups have designed hardware architectures for modeling plasticity in SNNs for various applications. Following these research efforts, this paper proposes multiplier‐less digital neuromorphic circuits for two plasticity learning rules: the spike‐driven synaptic plasticity (SDSP) and synaptic strength–based spike timing–dependent plasticity (SSSTDP). The proposed architectures have increased the precision of the plastic synaptic weights and are suitable for spiking neural network architectures with more precise calculations. The proposed models are validated in MATLAB simulations and physical implementations on a field‐programmable gate array (FPGA).

Digital Implementation of a Spiking Convolutional Neural Network for Tumor Detection

In this paper, an architecture for a Spiking Convolutional Neural Networks (SCNNs) has been implemented in an embedded system. The aim of this implementation is to present the ability of CNNs in order to hardware utilization and power consumption in complex applications such as tumor detection. Accordingly, the structure of the proposed SCNN is deployed on an FPGA by using fixed point arithmetic. The structural variation of the brain tissue creates challenges for detection of tumors in MRI images. To evaluate the speed, accuracy, and flexibility of the proposed SCNN, Izhikevich neuron model with the spike-timing-dependent plasticity (STDP) learning rule is used. The suggested neural network is explored, considering digital implementation possibility and costs. Results of the hardware synthesis and digital implementation on a field-programmable gate array (FPGA) are presented as a proof on concept.

Controlled Forgetting: Targeted Stimulation and Dopaminergic Plasticity Modulation for Unsupervised Lifelong Learning in Spiking Neural Networks.

Stochastic gradient descent requires that training samples be drawn from a uniformly random distribution of the data. For a deployed system that must learn online from an uncontrolled and unknown environment, the ordering of input samples often fails to meet this criterion, making lifelong learning a difficult challenge. We exploit the locality of the unsupervised Spike Timing Dependent Plasticity (STDP) learning rule to target local representations in a Spiking Neural Network (SNN) to adapt to novel information while protecting essential information in the remainder of the SNN from catastrophic forgetting. In our Controlled Forgetting Networks (CFNs), novel information triggers stimulated firing and heterogeneously modulated plasticity, inspired by biological dopamine signals, to cause rapid and isolated adaptation in the synapses of neurons associated with outlier information. This targeting controls the forgetting process in a way that reduces the degradation of accuracy for older tasks while learning new tasks. Our experimental results on the MNIST dataset validate the capability of CFNs to learn successfully over time from an unknown, changing environment, achieving 95.24% accuracy, which we believe is the best unsupervised accuracy ever achieved by a fixed-size, single-layer SNN on a completely disjoint MNIST dataset.

Realization of Synapse Behaviors Based on Memristor and Simulation Study With KMC Method

The memristor, emulating biological synapse, is recognized as one key way to overcome the classic von Neumann bottleneck. In this work, by using active metal Cu as electrode, the device of Cu/GeTeOx/TiN exhibited typical resistive switching characteristic based on the electrochemical metallization mechanism (ECM). Moreover, it realized gradual potentiating and depressing conduction under DC and AC modes, which lead to emulating excitation and inhibition of biological synapses. According to these properties, spike-timing-dependent plasticity (STDP) learning rule was reproduced by applying appropriate pulse sequences. Furthermore, the kinetic Monte Carlo method was utilized to analyze and provide further demonstration for the behaviors of gradual change. These indicated that the device had great potential applied in the neuromorphic computing systems and the properties could be commonly realized by ECM mechanism.

Classifying Melanoma Skin Lesions Using Convolutional Spiking Neural Networks With Unsupervised STDP Learning Rule

Deep learning methods have made some achievements in the automatic skin lesion recognition, but there are still some problems such as limited training samples, too complicated network structure, and expensive computational costs. Considering the inherent power-efficiency, biological plausibility and good image recognition performance of spiking neural networks (SNNs), in this paper we make malignant melanoma and benign melanocytic nevi skin lesions classification using convolutional SNNs with unsupervised spike-timing-dependent plasticity (STDP) learning rule. Efficient temporal coding, event driven learning rule and winner-take-all (WTA) mechanism together ensure sparse spike coding and efficient learning of our networks which achieve an average accuracy of 83.8%. We further propose to use feature selection to select more diagnostic features to improve the classification performance of our networks. Our SNNs with feature selection reach an average accuracy of 87.7%. Experimental results show that comparing to CNNs that need to be trained from scratch, our SNNs (with and without feature selection) not only achieve much better classification accuracies but also have much better runtime efficiency. Moreover, although the pretrained CNNs models can achieve similar running time, our proposed SNNs are more stable and easier to use than the pretrained CNNs because we do not need to try many pretrained models any more, and our SNNs also have much better classification accuracies than the pretrained CNNs. In addition, our networks have only three convolutional layers, and the complexity of the model and the parameters that need to be trained in the networks are greatly reduced. Our works show that STDP-based SNNs are very beneficial for the implementation of automated skin lesion classifiers on small portable devices.

A carbon-based memristor design for associative learning activities and neuromorphic computing.

Carbon quantum dots (QDs) have attracted significant interest due to their excellent electronic properties and wide application prospects. However, the application of carbon QDs has been rarely reported in memristors. Here, a memristor model with carbon conductive filaments (CFs) is proposed for the first time based on carbon quantum dots. The CF-based devices exhibited excellent resistive switching performance, in particular a narrow range of SET and RESET voltages and good power efficiency and retention properties. These devices could also emulate important biological synapse performances, such as the transition from short-term plasticity (STP) to long-term potentiation (LTP) behaviors, long-term depression (LTD) behavior, and four types of spike-timing-dependent plasticity (STDP) learning rules. Interestingly, Pavlovian associative learning functions were also reliably demonstrated in the memristor device (MD). The digit recognition ability of the MDs was evaluated though a single-layer perceptron model, in which the recognition accuracy of digits reached 92.63% after 250 training iterations. The transmission electron microscopy (TEM) results evidenced that the carbon CF was found in the MD at the "ON" state. Thus, this new carbon CF-based mechanism for memristors provides a new idea for achieving better neuromorphic MDs and applications.

An Imbalanced R-STDP Learning Rule in Spiking Neural Networks for Medical Image Classification

Spiking neural networks (SNNs) have the advantages of inherent power-efficiency, biological plausibility and good image recognition performance. They are good candidates for medical image classification especially when the labeled training data are limited. In medical image classification, one of the major challenges is the highly class imbalanced problem which causes deep learning networks to bias towards the majority class and poorly recognizes the minority class. Despite that there are some methods for addressing this problem, very few algorithm-level methods exist for SNNs. In this work, we propose an imbalanced reward-modulation spike-timing-dependent plasticity (R-STDP) learning rule for SNNs to solve the medical image class imbalanced problem. We introduce an imbalanced reward coefficient for the R-STDP learning rule to set the reward from the minority class to be higher than that of the majority class, and this reward coefficient can help to set the class-dependent rewards according to the data statistic of the training dataset. Experiment results on three benchmark datasets with imbalanced splits show that our method significantly improves the performance than that of the baseline SNNs and outperforms the compared state-of-the-art methods addressing medical image class imbalanced problem including data-level and algorithm-level methods. Moreover, our method achieves excellent classification performance on the imbalanced medical dataset ISIC-2018. The results show that the proposed method can well help SNNs in classifying imbalanced medical image datasets. Besides, our proposed method can obtain high sensitivity to disease class by adjusting the reward coefficient, which is very useful for identifying disease samples in medical diagnostic tasks.

Optimal Distribution of Spiking Neurons Over Multicore Neuromorphic Processors

In a multicore neuromorphic processor embedding a learning algorithm, a presynaptic neuron is occasionally located in a different core from the cores of its postsynaptic neurons, which needs neuron-to-target core communication for inference through a network router. The more neuron-to-target core connections, the more workload is imposed on the network router, which the more likely causes event routing congestion. Another significant challenge arising from a large number of neuron-to-core connections is data duplication in multiple cores for the learning algorithm to access the full data to evaluate weight update. This data duplication consumes a considerable amount of on-chip memory while the memory capacity per core is strictly limited. The optimal distribution of neurons over cores is categorized as an optimization problem with constraints, which may allow the discrete Lagrangian multiplier method (LMM) to optimize the distribution. Proof-of-concept demonstrations were made on the distribution of neurons over cores in a neuromorphic processor embedding a learning algorithm. The choice of the learning algorithm was twofold: a simple spike timing-dependent plasticity learning rule and event-driven random backpropagation algorithm, which are categorized as a two- and three-factor learning rule, respectively. As a result, the discrete LMM significantly reduced the number of neuron-to-core connections for both algorithms by approximately 55% in comparison with the number for random distribution cases, implying a 55% reduction in the workload on the network router and a 52.8% reduction in data duplication. The code is available on-line (https://github.com/guhyunkim/Optimize-neuron-distribution).

Initial synaptic weight distribution for fast learning speed and high recognition rate in STDP-based spiking neural network

We analyze that the initial synaptic weight distribution affects the performance, such as the learning speed, recognition rate and the power consumption in the spiking neural networks (SNNs) based on spike-timing-dependent plasticity (STDP) learning rule. A thin-film transistor (TFT)-type NOR flash memory is used as a synaptic device. In this fully connected two-layer neuromorphic system using the proposed pulse scheme, the results with and without the homeostasis functionality were analyzed separately. In addition, power consumption of the network in various initial synaptic weight distributions, and recognition rate that varies with the number of output neurons are also investigated. In pattern recognition for 28 × 28 MNIST handwritten patterns, higher performance is achieved in various aspects when the initial synaptic weights are distributed near the maximum value.

Transposable 3T-SRAM Synaptic Array Using Independent Double-Gate Feedback Field-Effect Transistors

In this article, we present a transposable three-transistor static random access memory (3T-SRAM) array consisting of independent double-gate feedback field-effect transistors as binary synaptic devices and access transistors. The synaptic functions of the ${2} \times {2}$ SRAM array are investigated through mixed-mode technology computer-aided design simulations. This 3T-SRAM array provides parallel and bidirectional synaptic updates with fast operating speed. Furthermore, a simplified spike-timing-dependent plasticity learning rule is implemented by adjusting the widths of memory pulses. A compact cell area and a low-leakage power consumption allow this 3T-SRAM array to be used for adaptive synaptic devices in a large-scale neuromorphic system.

Electronic synapses with near-linear weight update using MoS2/graphene memristors

Emulating the human brain's circuitry composed of neurons and synapses is an emerging area of research in mitigating the “von Neumann bottleneck” in present computer architectures. The building block of these neuromorphic systems—the synapse—is commonly realized with oxide-based or phase change material-based devices, whose operation is limited by high programming currents and high reset currents. In this work, we have realized nonvolatile resistive switching MoS2/graphene devices that exhibit multiple conductance states at low operating currents. The MoS2/graphene devices exhibit essential synaptic behaviors, such as short and long-term potentiation, long-term depression, and the spike timing dependent plasticity learning rule. Most importantly, they exhibit a near-linear synaptic weight update, without any abrupt reset process, allowing their use in unsupervised learning applications. These electronic synapses are built with chemical vapor deposited MoS2 and graphene, demonstrating potential for large-scale realizations of machine learning hardware.

Deep Spiking Neural Network for Video-Based Disguise Face Recognition Based on Dynamic Facial Movements.

With the increasing popularity of social media and smart devices, the face as one of the key biometrics becomes vital for person identification. Among those face recognition algorithms, video-based face recognition methods could make use of both temporal and spatial information just as humans do to achieve better classification performance. However, they cannot identify individuals when certain key facial areas, such as eyes or nose, are disguised by heavy makeup or rubber/digital masks. To this end, we propose a novel deep spiking neural network architecture in this paper. It takes dynamic facial movements, the facial muscle changes induced by speaking or other activities, as the sole input. An event-driven continuous spike-timing-dependent plasticity learning rule with adaptive thresholding is applied to train the synaptic weights. The experiments on our proposed video-based disguise face database (MakeFace DB) demonstrate that the proposed learning method performs very well, i.e., it achieves from 95% to 100% correct classification rates under various realistic experimental scenarios.