Visual input underpins cognitive function by providing the brain with essential environmental information. Neural decoding of visual scenes seeks to reconstruct pixel-level images from neural activity, a vital capability for vision restoration via brain-computer interfaces. However, extracting visual content from time-resolved spiking activity remains a significant challenge. Here, we introduce the Wavelet-Informed Spike Augmentation (WISA) model, which applies multilevel wavelet transforms to spike trains to learn compact representations that can be directly fed into deep reconstruction networks. When tested on recorded retinal spike data responding to natural video stimuli, WISA substantially improves reconstruction accuracy, especially in recovering fine-grained details. These results emphasize the value of temporal spike patterns for high-fidelity visual decoding and demonstrate WISA as a promising model for visual decoding.

A single neuron receives an extensive array of synaptic inputs through its dendrites, raising the fundamental question of how these inputs undergo integration and summation, culminating in the initiation of spikes in the soma. Experimental and computational investigations have revealed various modes of integration operations that include linear, superlinear, and sublinear summation. Interestingly, different types of neurons exhibit diverse patterns of dendritic integration depending on the spatial distribution of dendrites. The functional implications of these specific integration modalities remain largely unexplored. In this study, we employ the Purkinje cell (PC) as a model system to investigate these complex questions. Our findings reveal that PCs generally exhibit sublinear summation across their expansive dendrites. Both spatial and temporal input dynamically modulates the degree of sublinearity. Strong sublinearity necessitates the synaptic distribution in PCs to be globally scattered sensitive, whereas weak sublinearity facilitates the generation of complex firing patterns in PCs. Using dendritic branches characterized by strong sublinearity as computational units, we demonstrate that a neuron can successfully address the feature binding problem. Taken together, these results offer a systematic perspective on the functional role of dendritic sublinearity, inspiring a broader understanding of dendritic integration in various neuronal types.

Recent studies in reinforcement learning have explored brain-inspired function approximators and learning algorithms to simulate brain intelligence and adapt to neuromorphic hardware. Among these approaches, reward-modulated spike-timing-dependent plasticity (R-STDP) is biologically plausible and energy-efficient, but suffers from a gap between its local learning rules and the global learning objectives, which limits its performance and applicability. In this paper, we design a recurrent winner-take-all network and propose the spiking variational policy gradient (SVPG), a new R-STDP learning method derived theoretically from the global policy gradient. Specifically, the policy inference is derived from an energy-based policy function using mean-field inference, and the policy optimization is based on a last-step approximation of the global policy gradient. These fill the gap between the local learning rules and the global target. In experiments including a challenging ViZDoom vision-based navigation task and two realistic robot control tasks, SVPG successfully solves all the tasks. In addition, SVPG exhibits better inherent robustness to various kinds of input, network parameters, and environmental perturbations than compared methods.

Predicting the binding affinity of drug target is essential to reduce drug development costs and cycles. Recently, several deep learning-based methods have been proposed to utilize the structural or sequential information of drugs and targets to predict the drug-target binding affinity (DTA). However, methods that rely solely on sequence features do not consider hydrogen atom data, which may result in information loss. Graph-based methods may contain information that is not directly related to the prediction process. Additionally, the lack of structured division can limit the representation of characteristics. To address these issues, we propose a multimodal DTA prediction model using graph local substructures, called MLSDTA. This model comprehensively integrates the graph and sequence modal information from drugs and targets, achieving multimodal fusion through a cross-attention approach for multimodal features. Additionally, adaptive structure aware pooling is applied to generate graphs containing local substructural information. The model also utilizes the DropNode strategy to enhance the distinctions between different molecules. Experiments on two benchmark datasets have shown that MLSDTA outperforms current state-of-the-art models, demonstrating the feasibility of MLSDTA.

Spiking neural networks (SNNs) exhibit significant advantages in terms of information encoding, computational capabilities, and power usage. We regard initializing weight distribution as a key problem for effective SNN training. When backpropagation (BP) through time is used in the initial training phase, it has a significant impact on gradient generation. We first derive an asymptotic formula for the response curve of spiking neurons, which approximates the real neuron response distribution. To avoid gradient vanishing, we then provide an initialization technique based on the slant asymptote. Finally, validations on classification tasks on the MNIST and CIFAR10 datasets demonstrate that our strategy can significantly speed up training and improve the model accuracy compared with other initialization methods. Further testing on various neuron configurations and training hyperparameters demonstrates comparable versatility and superiority to other methods. Based on the analyses, some recommendations for SNN training are made.

Spiking neural networks (SNNs) have garnered interest due to their energy efficiency and superior effectiveness on neuromorphic chips compared with traditional artificial neural networks (ANNs). One of the mainstream approaches to implementing deep SNNs is the ANN-SNN conversion, which integrates the efficient training strategy of ANNs with the energy-saving potential and fast inference capability of SNNs. However, under extremely low-latency conditions, the existing conversion theory suggests that the problem of SNNs' neurons firing more or fewer spikes within each layer, i.e., residual error, leads to a performance gap in the converted SNNs compared with the original ANNs. This severely limits the possibility of the practical application of SNNs on delay-sensitive edge devices. Existing conversion methods addressing this problem usually involve modifying the state of the conversion spiking neurons. However, these methods do not consider their adaptability and compatibility with neuromorphic chips. We propose a new approach based on explicit modeling of residual errors as additive noise. The noise is incorporated into the activation function of the source ANN, effectively reducing the impact of residual error on SNN performance. Our experiments on the CIFAR10/100 and Tiny-ImageNet datasets verify that our approach exceeds the prevailing ANN-SNN conversion methods and directly trained SNNs concerning accuracy and the required time steps. Overall, our method provides new ideas for improving SNN performance under ultralow-latency conditions and is expected to promote practical neuromorphic hardware applications for further development. The code for our NQ framework is available at https://github.com/hzp2022/ANN2SNN_NQ.

Sensory information recognition is primarily processed through the ventral and dorsal visual pathways in the primate brain visual system, which exhibits layered feature representations bearing a strong resemblance to convolutional neural networks (CNNs), encompassing reconstruction and classification. However, existing studies often treat these pathways as distinct entities, focusing individually on pattern reconstruction or classification tasks, overlooking a key feature of biological neurons, the fundamental units for neural computation of visual sensory information. Addressing these limitations, we introduce a unified framework for sensory information recognition with augmented spikes. By integrating pattern reconstruction and classification within a single framework, our approach not only accurately reconstructs multimodal sensory information but also provides precise classification through definitive labeling. Experimental evaluations conducted on various datasets including video scenes, static images, dynamic auditory scenes, and functional magnetic resonance imaging (fMRI) brain activities demonstrate that our framework delivers state-of-the-art pattern reconstruction quality and classification accuracy. The proposed framework enhances the biological realism of multimodal pattern recognition models, offering insights into how the primate brain visual system effectively accomplishes the reconstruction and classification tasks through the integration of ventral and dorsal pathways.

Sensory information transmitted to the brain activates neurons to create a series of coping behaviors. Understanding the mechanisms of neural computation and reverse engineering the brain to build intelligent machines requires establishing a robust relationship between stimuli and neural responses. Neural decoding aims to reconstruct the original stimuli that trigger neural responses. With the recent upsurge of artificial intelligence, neural decoding provides an insightful perspective for designing novel algorithms of brain-machine interface. For humans, vision is the dominant contributor to the interaction between the external environment and the brain. In this study, utilizing the retinal neural spike data collected over multi trials with visual stimuli of two movies with different levels of scene complexity, we used a neural network decoder to quantify the decoded visual stimuli with six different metrics for image quality assessment establishing comprehensive inspection of decoding. With the detailed and systematical study of the effect and single and multiple trials of data, different noise in spikes, and blurred images, our results provide an in-depth investigation of decoding dynamical visual scenes using retinal spikes. These results provide insights into the neural coding of visual scenes and services as a guideline for designing next-generation decoding algorithms of neuroprosthesis and other devices of brain-machine interface.

Spiking neural networks (SNNs) exploit neural spikes to provide solutions for low-power intelligent applications on neuromorphic hardware. Although SNNs have high computational efficiency due to spiking communication, they still lack resistance to adversarial attacks and noise perturbations. In the brain, neuronal responses generally possess stochasticity induced by ion channels and synapses, while the role of stochasticity in computing tasks is poorly understood. Inspired by this, we elaborate a stochastic gating spiking neural model for layer-by-layer spike communication, introducing stochasticity to SNNs. Through theoretical analysis, our gating model can be viewed as a regularizer that prevents error amplification under attacks. Meanwhile, our work can explain the robustness of Poisson coding. Experimental results prove that our method can be used alone or with existing robust enhancement algorithms to improve SNN robustness and reduce SNN energy consumption. We hope our work will shed new light on the role of stochasticity in the computation of SNNs. Our code is available at https://github.com/DingJianhao/StoG-meets-SNN/.

Optimizing multi-modal learning of SNNs has the advantages of energy efficiency and performance improvements. However, modality imbalance in multi-modal SNNs results in performance decline due to heterogeneity of modalities and temporal inconsistencies across different SNNs branches. In this paper, we propose the Balanced Multi-modal SNNs (BM-SNNs) model, equipped with a novel online loss adjustment (LA) algorithm and time alignment (TA) modules, ultimately achieving balanced training across multiple modalities. LA supervises the learning of uni-modal feature extractors by adding unimodal loss components without additional classifier. Moreover, the modulation factors enable the adaptive adjustment of unimodal learning rate. Furthermore, the proposed TA adopts the optimal timestep for different modalities to avoid information redundancy. Experimental results reveal that BM-SNNs model improves the performance of both multi-modal model and unimodal branches through adaptively exploiting intra-modal and cross-modal information.

Understanding the computational mechanisms that underlie the encoding and decoding of environmental stimuli is a crucial investigation in neuroscience. Central to this pursuit is the exploration of how the brain represents visual information across its hierarchical architecture. A prominent challenge resides in discerning the neural underpinnings of the processing of dynamic natural visual scenes. Although considerable research efforts have been made to characterize individual components of the visual pathway, a systematic understanding of the distinctive neural coding associated with visual stimuli, as they traverse this hierarchical landscape, remains elusive. In this study, we leverage the comprehensive Allen Visual Coding-Neuropixels dataset and utilize the capabilities of deep learning neural network models to study neural coding in response to dynamic natural visual scenes across an expansive array of brain regions. Our study reveals that our decoding model adeptly deciphers visual scenes from neural spiking patterns exhibited within each distinct brain area. A compelling observation arises from the comparative analysis of decoding performances, which manifests as a notable encoding proficiency within the visual cortex and subcortical nuclei, in contrast to a relatively reduced encoding activity within hippocampal neurons. Strikingly, our results unveil a robust correlation between our decoding metrics and well-established anatomical and functional hierarchy indexes. These findings corroborate existing knowledge in visual coding related to artificial visual stimuli and illuminate the functional role of these deeper brain regions using dynamic stimuli. Consequently, our results suggest a novel perspective on the utility of decoding neural network models as a metric for quantifying the encoding quality of dynamic natural visual scenes represented by neural responses, thereby advancing our comprehension of visual coding within the complex hierarchy of the brain.

Representation learning in heterogeneous graphs with massive unlabeled data has aroused great interest. The heterogeneity of graphs not only contains rich information, but also raises difficult barriers to designing unsupervised or self-supervised learning (SSL) strategies. Existing methods such as random walk-based approaches are mainly dependent on the proximity information of neighbors and lack the ability to integrate node features into a higher-level representation. Furthermore, previous self-supervised or unsupervised frameworks are usually designed for node-level tasks, which are commonly short of capturing global graph properties and may not perform well in graph-level tasks. Therefore, a label-free framework that can better capture the global properties of heterogeneous graphs is urgently required. In this article, we propose a self-supervised heterogeneous graph neural network (GNN) based on cross-view contrastive learning (HeGCL). The HeGCL presents two views for encoding heterogeneous graphs: the meta-path view and the outline view. Compared with the meta-path view that provides semantic information, the outline view encodes the complex edge relations and captures graph-level properties by using a nonlocal block. Thus, the HeGCL learns node embeddings through maximizing mutual information (MI) between global and semantic representations coming from the outline and meta-path view, respectively. Experiments on both node-level and graph-level tasks show the superiority of the proposed model over other methods, and further exploration studies also show that the introduction of nonlocal block brings a significant contribution to graph-level tasks.

Thanks to their event-driven nature, spiking neural networks (SNNs) are surmised to be great computation-efficient models. The spiking neurons encode beneficial temporal facts and possess excessive anti-noise properties. However, the high-quality encoding of spatio-temporal complexity and also its training optimization of SNNs are restricted by means of the contemporary problem, this article proposes a novel hierarchical event-driven visual device to explore how information transmits and signifies in the retina the usage of biologically manageable mechanisms. This cognitive model is an augmented spiking-based framework consisting of the function learning capacity of convolutional neural networks (CNNs) with the cognition capability of SNNs. Furthermore, this visual device is modeled in a biological realism way with unsupervised learning rules and advanced spike firing rate encoding methods. We train and test them on some image datasets (Modified National Institute of Standards and Technology (MNIST), Canadian Institute for Advanced Research (CIFAR)10, and its noisy versions) to show that our mannequin can process greater vital data than present cognitive models. This article also proposes a novel quantization approach to make the proposed spiking-based model more efficient for neuromorphic hardware implementation. The outcomes show this joint CNN-SNN model can reap excessive focus accuracy and get more effective generalization ability.

Spiking neural networks (SNNs) have manifested remarkable advantages in power consumption and event-driven property during the inference process. To take full advantage of low power consumption and improve the efficiency of these models further, the pruning methods have been explored to find sparse SNNs without redundancy connections after training. However, parameter redundancy still hinders the efficiency of SNNs during training. In the human brain, the rewiring process of neural networks is highly dynamic, while synaptic connections maintain relatively sparse during brain development. Inspired by this, here we propose an efficient evolutionary structure learning (ESL) framework for SNNs, named ESL-SNNs, to implement the sparse SNN training from scratch. The pruning and regeneration of synaptic connections in SNNs evolve dynamically during learning, yet keep the structural sparsity at a certain level. As a result, the ESL-SNNs can search for optimal sparse connectivity by exploring all possible parameters across time. Our experiments show that the proposed ESL-SNNs framework is able to learn SNNs with sparse structures effectively while reducing the limited accuracy. The ESL-SNNs achieve merely 0.28% accuracy loss with 10% connection density on the DVS-Cifar10 dataset. Our work presents a brand-new approach for sparse training of SNNs from scratch with biologically plausible evolutionary mechanisms, closing the gap in the expressibility between sparse training and dense training. Hence, it has great potential for SNN lightweight training and inference with low power consumption and small memory usage.

Spiking neural networks (SNNs) are well-known as brain-inspired models with high computing efficiency, due to a key component that they utilize spikes as information units, close to the biological neural systems. Although spiking based models are energy efficient by taking advantage of discrete spike signals, their performance is limited by current network structures and their training methods. As discrete signals, typical SNNs cannot apply the gradient descent rules directly into parameter adjustment as artificial neural networks (ANNs). Aiming at this limitation, here we propose a novel method of constructing deep SNN models with knowledge distillation (KD) that uses ANN as the teacher model and SNN as the student model. Through the ANN-SNN joint training algorithm, the student SNN model can learn rich feature information from the teacher ANN model through the KD method, yet it avoids training SNN from scratch when communicating with non-differentiable spikes. Our method can not only build a more efficient deep spiking structure feasibly and reasonably but use few time steps to train the whole model compared to direct training or ANN to SNN methods. More importantly, it has a superb ability of noise immunity for various types of artificial noises and natural signals. The proposed novel method provides efficient ways to improve the performance of SNN through constructing deeper structures in a high-throughput fashion, with potential usage for light and efficient brain-inspired computing of practical scenarios.

Neurons, represented as a tree structure of morphology, have various distinguished branches of dendrites. Different types of synaptic receptors distributed over dendrites are responsible for receiving inputs from other neurons. NMDA receptors (NMDARs) are expressed as excitatory units, and play a key physiological role in synaptic function. Although NMDARs are widely expressed in most types of neurons, they play a different role in the cerebellar Purkinje cells (PCs). Utilizing a computational PC model with detailed dendritic morphology, we explored the role of NMDARs at different parts of dendritic branches and regions. We found somatic responses can switch from silent, to simple spikes and complex spikes, depending on specific dendritic branches. Detailed examination of the dendrites regarding their diameters and distance to soma revealed diverse response patterns, yet explain two firing modes, simple and complex spike. Taken together, these results suggest that NMDARs play an important role in controlling excitability sensitivity while taking into account the factor of dendritic properties. Given the complexity of neural morphology varying in cell types, our work suggests that the functional role of NMDARs is not stereotyped but highly interwoven with local properties of neuronal structure.

Spiking neural networks (SNNs), inspired by the neuronal network in the brain, provide biologically relevant and low-power consuming models for information processing. Existing studies either mimic the learning mechanism of brain neural networks as closely as possible, for example, the temporally local learning rule of spike-timing-dependent plasticity (STDP), or apply the gradient descent rule to optimize a multilayer SNN with fixed structure. However, the learning rule used in the former is local and how the real brain might do the global-scale credit assignment is still not clear, which means that those shallow SNNs are robust but deep SNNs are difficult to be trained globally and could not work so well. For the latter, the nondifferentiable problem caused by the discrete spike trains leads to inaccuracy in gradient computing and difficulties in effective deep SNNs. Hence, a hybrid solution is interesting to combine shallow SNNs with an appropriate machine learning (ML) technique not requiring the gradient computing, which is able to provide both energy-saving and high-performance advantages. In this article, we propose a HybridSNN, a deep and strong SNN composed of multiple simple SNNs, in which data-driven greedy optimization is used to build powerful classifiers, avoiding the derivative problem in gradient descent. During the training process, the output features (spikes) of selected weak classifiers are fed back to the pool for the subsequent weak SNN training and selection. This guarantees HybridSNN not only represents the linear combination of simple SNNs, as what regular AdaBoost algorithm generates, but also contains neuron connection information, thus closely resembling the neural networks of a brain. HybridSNN has the benefits of both low power consumption in weak units and overall data-driven optimizing strength. The network structure in HybridSNN is learned from training samples, which is more flexible and effective compared with existing fixed multilayer SNNs. Moreover, the topological tree of HybridSNN resembles the neural system in the brain, where pyramidal neurons receive thousands of synaptic input signals through their dendrites. Experimental results show that the proposed HybridSNN is highly competitive among the state-of-the-art SNNs.

Spiking neural networks (SNNs) are promising to be widely deployed in real-time and safety-critical applications with the advance of neuromorphic computing. Recent work has demonstrated the insensitivity of SNNs to small random perturbations due to the discrete internal information representation. The variety of training algorithms and the involvement of the temporal dimension pose more threats to the robustness of SNNs than that of typical neural networks. We account for the vulnerability of SNNs by constructing adversaries based on different differentiable approximation techniques. By deriving a Lipschitz constant specifically for the spike representation, we first theoretically answer the question of how much adversarial invulnerability is retained in SNNs. Hence, to defend against the broad attack methods, we propose a regularized adversarial training scheme with low computational overheads. SNNs can benefit from the constraint of the perturbed spike distance's amplification and the generalization on multiple adversarial $\epsilon$-neighbourhoods. Our experiments on the image recognition benchmarks have proven that our training scheme can defend against powerful adversarial attacks crafted from strong differentiable approximations. To be specific, our approach makes the black-box attacks of the Projected Gradient Descent attack nearly ineffective. We believe that our work will facilitate the spread of SNNs for safety-critical applications and help understand the robustness of the human brain.

Images of visual scenes comprise essential features important for visual cognition of the brain. The complexity of visual features lies at different levels, from simple artificial patterns to natural images with different scenes. It has been a focus of using stimulus images to predict neural responses. However, it remains unclear how to extract features from neuronal responses. Here we address this question by leveraging two-photon calcium neural data recorded from the visual cortex of awake macaque monkeys. With stimuli including various categories of artificial patterns and diverse scenes of natural images, we employed a deep neural network decoder inspired by image segmentation technique. Consistent with the notation of sparse coding for natural images, a few neurons with stronger responses dominated the decoding performance, whereas decoding of ar tificial patterns needs a large number of neurons. When natural images using the model pretrained on artificial patterns are decoded, salient features of natural scenes can be extracted, as well as the conventional category information. Altogether, our results give a new perspective on studying neural encoding principles using reverse-engineering decoding strategies.

A central goal in sensory neuroscience is to understand the neuronal signal processing involved in the encoding of natural stimuli. A critical step towards this goal is the development of successful computational encoding models. For ganglion cells in the vertebrate retina, the development of satisfactory models for responses to natural visual scenes is an ongoing challenge. Standard models typically apply linear integration of visual stimuli over space, yet many ganglion cells are known to show nonlinear spatial integration, in particular when stimulated with contrast-reversing gratings. We here study the influence of spatial nonlinearities in the encoding of natural images by ganglion cells, using multielectrode-array recordings from isolated salamander and mouse retinas. We assess how responses to natural images depend on first- and second-order statistics of spatial patterns inside the receptive field. This leads us to a simple extension of current standard ganglion cell models. We show that taking not only the weighted average of light intensity inside the receptive field into account but also its variance over space can partly account for nonlinear integration and substantially improve response predictions of responses to novel images. For salamander ganglion cells, we find that response predictions for cell classes with large receptive fields profit most from including spatial contrast information. Finally, we demonstrate how this model framework can be used to assess the spatial scale of nonlinear integration. Our results underscore that nonlinear spatial stimulus integration translates to stimulation with natural images. Furthermore, the introduced model framework provides a simple, yet powerful extension of standard models and may serve as a benchmark for the development of more detailed models of the nonlinear structure of receptive fields.

Neuronal circuits formed in the brain are complex with intricate connection patterns. Such complexity is also observed in the retina with a relatively simple neuronal circuit. A retinal ganglion cell (GC) receives excitatory inputs from neurons in previous layers as driving forces to fire spikes. Analytical methods are required to decipher these components in a systematic manner. Recently a method called spike-triggered non-negative matrix factorization (STNMF) has been proposed for this purpose. In this study, we extend the scope of the STNMF method. By using retinal GCs as a model system, we show that STNMF can detect various computational properties of upstream bipolar cells (BCs), including spatial receptive field, temporal filter, and transfer nonlinearity. In addition, we recover synaptic connection strengths from the weight matrix of STNMF. Furthermore, we show that STNMF can separate spikes of a GC into a few subsets of spikes, where each subset is contributed by one presynaptic BC. Taken together, these results corroborate that STNMF is a useful method for deciphering the structure of neuronal circuits.

Deep convolutional neural networks (CNNs) have demonstrated impressive performance on many visual tasks. Recently, they became useful models for the visual system in neuroscience. However, it is still not clear what is learned by CNNs in terms of neuronal circuits. When a deep CNN with many layers is used for the visual system, it is not easy to compare the structure components of CNNs with possible neuroscience underpinnings due to highly complex circuits from the retina to the higher visual cortex. Here, we address this issue by focusing on single retinal ganglion cells with biophysical models and recording data from animals. By training CNNs with white noise images to predict neuronal responses, we found that fine structures of the retinal receptive field can be revealed. Specifically, convolutional filters learned are resembling biological components of the retinal circuit. This suggests that a CNN learning from one single retinal cell reveals a minimal neural network carried out in this cell. Furthermore, when CNNs learned from different cells are transferred between cells, there is a diversity of transfer learning performance, which indicates that CNNs are cell specific. Moreover, when CNNs are transferred between different types of input images, here white noise versus natural images, transfer learning shows a good performance, which implies that CNNs indeed capture the full computational ability of a single retinal cell for different inputs. Taken together, these results suggest that CNNs could be used to reveal structure components of neuronal circuits, and provide a powerful model for neural system identification.

Neural coding, including encoding and decoding, is one of the key problems in neuroscience for understanding how the brain uses neural signals to relate sensory perception and motor behaviors with neural systems. However, most of the existed studies only aim at dealing with the continuous signal of neural systems, while lacking a unique feature of biological neurons, termed spike, which is the fundamental information unit for neural computation as well as a building block for brain-machine interface. Aiming at these limitations, we propose a transcoding framework to encode multi-modal sensory information into neural spikes and then reconstruct stimuli from spikes. Sensory information can be compressed into 10% in terms of neural spikes, yet re-extract 100% of information by reconstruction. Our framework can not only feasibly and accurately reconstruct dynamical visual and auditory scenes, but also rebuild the stimulus patterns from functional magnetic resonance imaging (fMRI) brain activities. More importantly, it has a superb ability of noise immunity for various types of artificial noises and background signals. The proposed framework provides efficient ways to perform multimodal feature representation and reconstruction in a high-throughput fashion, with potential usage for efficient neuromorphic computing in a noisy environment.

Our real-time actions in everyday life reflect a range of spatiotemporal dynamic brain activity patterns, the consequence of neuronal computation with spikes in the brain. Most existing models with spiking neurons aim at solving static pattern recognition tasks such as image classification. Compared with static features, spatiotemporal patterns are more complex due to their dynamics in both space and time domains. Spatiotemporal pattern recognition based on learning algorithms with spiking neurons therefore remains challenging. We propose an end-to-end recurrent spiking neural network model trained with an algorithm based on spike latency and temporal difference backpropagation. Our model is a cascaded network with three layers of spiking neurons where the input and output layers are the encoder and decoder, respectively. In the hidden layer, the recurrently connected neurons with transmission delays carry out high-dimensional computation to incorporate the spatiotemporal dynamics of the inputs. The test results based on the data sets of spiking activities of the retinal neurons show that the proposed framework can recognize dynamic spatiotemporal patterns much better than using spike counts. Moreover, for 3D trajectories of a human action data set, the proposed framework achieves a test accuracy of 83.6% on average. Rapid recognition is achieved through the learning methodology-based on spike latency and the decoding process using the first spike of the output neurons. Taken together, these results highlight a new model to extract information from activity patterns of neural computation in the brain and provide a novel approach for spike-based neuromorphic computing.

Finding out the physical structure of neuronal circuits that governs neuronal responses is an important goal for brain research. With fast advances for large-scale recording techniques, identification of a neuronal circuit with multiple neurons and stages or layers becomes possible and highly demanding. Although methods for mapping the connection structure of circuits have been greatly developed in recent years, they are mostly limited to simple scenarios of a few neurons in a pairwise fashion; and dissecting dynamical circuits, particularly mapping out a complete functional circuit that converges to a single neuron, is still a challenging question. Here, we show that a recent method, termed spike-triggered non-negative matrix factorization (STNMF), can address these issues. By simulating different scenarios of spiking neural networks with various connections between neurons and stages, we demonstrate that STNMF is a persuasive method to dissect functional connections within a circuit. Using spiking activities recorded at neurons of the output layer, STNMF can obtain a complete circuit consisting of all cascade computational components of presynaptic neurons, as well as their spiking activities. For simulated simple and complex cells of the primary visual cortex, STNMF allows us to dissect the pathway of visual computation. Taken together, these results suggest that STNMF could provide a useful approach for investigating neuronal systems leveraging recorded functional neuronal activity.

Synchronous oscillations in neural populations are considered being controlled by inhibitory neurons. In the granular layer of the cerebellum, two major types of cells are excitatory granular cells (GCs) and inhibitory Golgi cells (GoCs). GC spatiotemporal dynamics, as the output of the granular layer, is highly regulated by GoCs. However, there are various types of inhibition implemented by GoCs. With inputs from mossy fibers, GCs and GoCs are reciprocally connected to exhibit different network motifs of synaptic connections. From the view of GCs, feedforward inhibition is expressed as the direct input from GoCs excited by mossy fibers, whereas feedback inhibition is from GoCs via GCs themselves. In addition, there are abundant gap junctions between GoCs showing another form of inhibition. It remains unclear how these diverse copies of inhibition regulate neural population oscillation changes. Leveraging a computational model of the granular layer network, we addressed this question to examine the emergence and modulation of network oscillation using different types of inhibition. We show that at the network level, feedback inhibition is crucial to generate neural oscillation. When short-term plasticity was equipped on GoC-GC synapses, oscillations were largely diminished. Robust oscillations can only appear with additional gap junctions. Moreover, there was a substantial level of cross-frequency coupling in oscillation dynamics. Such a coupling was adjusted and strengthened by GoCs through feedback inhibition. Taken together, our results suggest that the cooperation of distinct types of GoC inhibition plays an essential role in regulating synchronous oscillations of the GC population. With GCs as the sole output of the granular network, their oscillation dynamics could potentially enhance the computational capability of downstream neurons.

The dynamics of cerebellar neuronal networks is controlled by the underlying building blocks of neurons and synapses between them. For which, the computation of Purkinje cells (PCs), the only output cells of the cerebellar cortex, is implemented through various types of neural pathways interactively routing excitation and inhibition converged to PCs. Such tuning of excitation and inhibition, coming from the gating of specific pathways as well as short-term plasticity (STP) of the synapses, plays a dominant role in controlling the PC dynamics in terms of firing rate and spike timing. PCs receive cascade feedforward inputs from two major neural pathways: the first one is the feedforward excitatory pathway from granule cells (GCs) to PCs; the second one is the feedforward inhibition pathway from GCs, via molecular layer interneurons (MLIs), to PCs. The GC-PC pathway, together with short-term dynamics of excitatory synapses, has been a focus over past decades, whereas recent experimental evidence shows that MLIs also greatly contribute to controlling PC activity. Therefore, it is expected that the diversity of excitation gated by STP of GC-PC synapses, modulated by strong inhibition from MLI-PC synapses, can promote the computation performed by PCs. However, it remains unclear how these two neural pathways are interacted to modulate PC dynamics. Here using a computational model of PC network installed with these two neural pathways, we addressed this question to investigate the change of PC firing dynamics at the level of single cell and network. We show that the nonlinear characteristics of excitatory STP dynamics can significantly modulate PC spiking dynamics mediated by inhibition. The changes in PC firing rate, firing phase, and temporal spike pattern, are strongly modulated by these two factors in different ways. MLIs mainly contribute to variable delays in the postsynaptic action potentials of PCs while modulated by excitation STP. Notably, the diversity of synchronization and pause response in the PC network is governed not only by the balance of excitation and inhibition, but also by the synaptic STP, depending on input burst patterns. Especially, the pause response shown in the PC network can only emerge with the interaction of both pathways. Together with other recent findings, our results show that the interaction of feedforward pathways of excitation and inhibition, incorporated with synaptic short-term dynamics, can dramatically regulate the PC activities that consequently change the network dynamics of the cerebellar circuit.

The automatic score detection and recognition in basketball game has important application potentials, for examples, basketball technique analysis and 24 second control in the game. Although existing studies have been conducted on broadcast videos, most of them usually learned a machine learning algorithm on long videos recorded by traditional cameras. Address Event Representation (AER) sensor provides a possibility to deal with the problem by a human sensing manner. It represents the visual information as a series of spike-based events and records event sequences. Compared to traditional videos, AER events can fully utilize their addresses and timestamp information, forming precise spatio-temporal features with significantly less storage cost. More importantly, it issues spikes which can be naturally processed by human-style spiking neural networks (SNNs). In this paper, we propose to recognize scoring in basketball game from AER sequences. A new model is designed to extract dynamic features and discriminate different event streams using SNN. To handle the imbalance problem between positive and negative samples, we use an imbalanced Tempotron algorithm in our SNN model. Meanwhile, an AER sequence dataset of basketball games is collected. The experimental results demonstrate that our method achieves better performance compared with existing models.

Neural encoding and decoding of retinal ganglion cells (RGCs) have been attached great importance in the research work of brain-machine interfaces. Much effort has been invested to mimic RGC and get insight into RGC signals to reconstruct stimuli. However, there remain two challenges. On the one hand, complex nonlinear processes in retinal neural circuits hinder encoding models from enhancing their ability to fit the natural stimuli and modelling RGCs accurately. On the other hand, current research of the decoding process is separate from that of the encoding process, in which the liaison of mutual promotion between them is neglected. In order to alleviate the above problems, we propose a cross-modal dual deep generative model (CDDG) in this paper. CDDG treats the RGC spike signals and the stimuli as two modalities, which learns a shared latent representation for the concatenated modality and two modal-specific latent representations. Then, it imposes distribution consistency restriction on different latent space, cross-consistency and cycle-consistency constraints on the generated variables. Thus, our model ensures cross-modal generation from RGC spike signals to stimuli and vice versa. In our framework, the generation from stimuli to RGC spike signals is equivalent to neural encoding while the inverse process is equivalent to neural decoding. Hence, the proposed method integrates neural encoding and decoding and exploits the reciprocity between them. The experimental results demonstrate that our proposed method can achieve excellent encoding and decoding performance compared with the state-of-the-art methods on three salamander RGC spike datasets with natural stimuli.

Hidden Markov models (HMMs) underpin the solution to many problems in computational neuroscience. However, it is still unclear how to implement inference of HMMs with a network of neurons in the brain. The existing methods suffer from the problem of being nonspiking and inaccurate. Here, we build a precise equivalence between the inference equation of HMMs with time-invariant hidden variables and the dynamics of spiking winner-take-all (WTA) neural networks. We show that the membrane potential of each spiking neuron in the WTA circuit encodes the logarithm of the posterior probability of the hidden variable in each state, and the firing rate of each neuron is proportional to the posterior probability of the HMMs. We prove that the time course of the neural firing rate can implement posterior inference of HMMs. Theoretical analysis and experimental results show that the proposed WTA circuit can get accurate inference results of HMMs.

Numerous neuroscience experiments have suggested that the cognitive process of human brain is realized as probability reasoning and further modeled as Bayesian inference. It is still unclear how Bayesian inference could be implemented by neural underpinnings in the brain. Here we present a novel Bayesian inference algorithm based on importance sampling. By distributed sampling through a deep tree structure with simple and stackable basic motifs for any given neural circuit, one can perform local inference while guaranteeing the accuracy of global inference. We show that these task-independent motifs can be used in parallel for fast inference without iteration and scale-limitation. Furthermore, experimental simulations with a small-scale neural network demonstrate that our distributed sampling-based algorithm, consisting with our theoretical analysis, can approximate Bayesian inference. Taken all together, we provide a proofof- principle to use distributed neural networks to implement Bayesian inference, which gives a road-map for large-scale Bayesian network implementation based on spiking neural networks with computer hardwares, including neuromorphic chips.

The processing capabilities of biological vision systems are still vastly superior to artificial vision, even though this has been an active area of research for over half a century. Current artificial vision techniques integrate many insights from biology yet they remain far-off the capabilities of animals and humans in terms of speed, power, and performance. A key aspect to modeling the human visual system is the ability to accurately model the behavior and computation within the retina. In particular, we focus on modeling the retinal ganglion cells (RGCs) as they convey the accumulated data of real world images as action potentials onto the visual cortex via the optic nerve. Computational models that approximate the processing that occurs within RGCs can be derived by quantitatively fitting the sets of physiological data using an input-output analysis where the input is a known stimulus and the output is neuronal recordings. Currently, these input-output responses are modeled using computational combinations of linear and nonlinear models that are generally complex and lack any relevance to the underlying biophysics. In this paper, we illustrate how system identification techniques, which take inspiration from biological systems, can accurately model retinal ganglion cell behavior, and are a viable alternative to traditional linear-nonlinear approaches.

Advances in neuronal recording techniques are leading to ever larger numbers of simultaneously monitored neurons. This poses the important analytical challenge of how to capture compactly all sensory information that neural population codes carry in their spatial dimension (differences in stimulus tuning across neurons at different locations), in their temporal dimension (temporal neural response variations), or in their combination (temporally coordinated neural population firing). Here we investigate the utility of tensor factorizations of population spike trains along space and time. These factorizations decompose a dataset of single-trial population spike trains into spatial firing patterns (combinations of neurons firing together), temporal firing patterns (temporal activation of these groups of neurons) and trial-dependent activation coefficients (strength of recruitment of such neural patterns on each trial). We validated various factorization methods on simulated data and on populations of ganglion cells simultaneously recorded in the salamander retina. We found that single-trial tensor space-by-time decompositions provided low-dimensional data-robust representations of spike trains that capture efficiently both their spatial and temporal information about sensory stimuli. Tensor decompositions with orthogonality constraints were the most efficient in extracting sensory information, whereas non-negative tensor decompositions worked well even on non-independent and overlapping spike patterns, and retrieved informative firing patterns expressed by the same population in response to novel stimuli. Our method showed that populations of retinal ganglion cells carried information in their spike timing on the ten-milliseconds-scale about spatial details of natural images. This information could not be recovered from the spike counts of these cells. First-spike latencies carried the majority of information provided by the whole spike train about fine-scale image features, and supplied almost as much information about coarse natural image features as firing rates. Together, these results highlight the importance of spike timing, and particularly of first-spike latencies, in retinal coding.

When visual contrast changes, retinal ganglion cells adapt by adjusting their sensitivity as well as their temporal filtering characteristics. The latter has classically been described by contrast-induced gain changes that depend on temporal frequency. Here, we explored a new perspective on contrast-induced changes in temporal filtering by using spike-triggered covariance analysis to extract multiple parallel temporal filters for individual ganglion cells. Based on multielectrode-array recordings from ganglion cells in the isolated salamander retina, we found that contrast adaptation of temporal filtering can largely be captured by contrast-invariant sets of filters with contrast-dependent weights. Moreover, differences among the ganglion cells in the filter sets and their contrast-dependent contributions allowed us to phenomenologically distinguish three types of filter changes. The first type is characterized by newly emerging features at higher contrast, which can be reproduced by computational models that contain response-triggered gain-control mechanisms. The second type follows from stronger adaptation in the Off pathway as compared to the On pathway in On-Off-type ganglion cells. Finally, we found that, in a subset of neurons, contrast-induced filter changes are governed by particularly strong spike-timing dynamics, in particular by pronounced stimulus-dependent latency shifts that can be observed in these cells. Together, our results show that the contrast dependence of temporal filtering in retinal ganglion cells has a multifaceted phenomenology and that a multi-filter analysis can provide a useful basis for capturing the underlying signal-processing dynamics.

It has been established that homeostatic synaptic scaling plasticity can maintain neural network activity in a stable regime. However, the underlying learning rule for this mechanism is still unclear. Whether it is dependent on the presynaptic site remains a topic of debate. Here we focus on two forms of learning rules: traditional synaptic scaling (SS) without presynaptic effect and presynaptic-dependent synaptic scaling (PSD). Analysis of the synaptic matrices reveals that transition matrices between consecutive synaptic matrices are distinct: they are diagonal and linear to neural activity under SS, but become nondiagonal and nonlinear under PSD. These differences produce different dynamics in recurrent neural networks. Numerical simulations show that network dynamics are stable under PSD but not SS, which suggests that PSD is a better form to describe homeostatic synaptic scaling plasticity. Matrix analysis used in the study may provide a novel way to examine the stability of learning dynamics.