Masked timesteps in spiking neural networks for efficient time series classification

For biomedical application, analysis of data through modern computational methodologies is required. Machine learning-based architectures enhance the way diagnosis is performed. The objective of this research work is to design a neuromorphic system using nanoelectronics and artificial intelligence for feature extraction and classification of medical data. The research area is a combination of nanoelectronics, computer technology, and biology. This research paper presents an extensive literature survey and descries the analysis, design, and implementation of neural networks based on human brain functionalities. The spiking neural networks (SNNs) architecture implementation in very large-scale integration will reduce the power consumption and miniaturize the device. A detailed review on various spiking neural networks architectures and methods is presented. In this research paper, an effort is made for the VLSI implementation of the spiking neural architecture. The implementation is carried out using Quartus tool and Spartan/Cyclone/Vertex Kits in 90 nm and 65 nm technology. Power, delay, and area are taken as the performance metrics.KeywordsFPGASpiking neural networkArtificial intelligenceNeuromorphic systemLow powerCMOSBiomedical

Developmental plasticity plays a prominent role in shaping the brain's structure during ongoing learning in response to dynamically changing environments. However, the existing network compression methods for deep artificial neural networks (ANNs) and spiking neural networks (SNNs) draw little inspiration from brain's developmental plasticity mechanisms, thus limiting their ability to learn efficiently, rapidly, and accurately. This paper proposed a developmental plasticity-inspired adaptive pruning (DPAP) method, with inspiration from the adaptive developmental pruning of dendritic spines, synapses, and neurons according to the "use it or lose it, gradually decay" principle. The proposed DPAP model considers multiple biologically realistic mechanisms (such as dendritic spine dynamic plasticity, activity-dependent neural spiking trace, and local synaptic plasticity), with additional adaptive pruning strategy, so that the network structure can be dynamically optimized during learning without any pre-training and retraining. Extensive comparative experiments show consistent and remarkable performance and speed boost with the extremely compressed networks on a diverse set of benchmark tasks for deep ANNs and SNNs, especially the spatio-temporal joint pruning of SNNs in neuromorphic datasets. This work explores how developmental plasticity enables complex deep networks to gradually evolve into brain-like efficient and compact structures, eventually achieving state-of-the-art (SOTA) performance for biologically realistic SNNs.

Spiking neural networks (SNNs) are quickly gaining traction as a viable alternative to deep neural networks (DNNs). Compared to DNNs, SNNs are computationally more powerful and energy efficient. The design metrics (synaptic weights, membrane threshold, etc.) chosen for such SNN architectures are often proprietary and constitute confidential intellectual property (IP). Our study indicates that SNN architectures implemented using conventional analog neurons are susceptible to side channel attack (SCA). Unlike the conventional SCAs that are aimed to leak private keys from cryptographic implementations, SCANN (SCA̲ of spiking n̲eural n̲etworks) can reveal the sensitive IP implemented within the SNN through the power side channel. We demonstrate eight unique SCANN attacks by taking a common analog neuron (axon hillock neuron) as the test case. We chose this particular model since it is biologically plausible and is hence a good fit for SNNs. Simulation results indicate that different synaptic weights, neurons/layer, neuron membrane thresholds, and neuron capacitor sizes (which are the building blocks of SNN) yield distinct power and spike timing signatures, making them vulnerable to SCA. We show that an adversary can use templates (using foundry-calibrated simulations or fabricating known design parameters in test chips) and analysis to identify the specifications of the implemented SNN.

Spiking neural networks (SNNs) have attracted many researchers’ interests due to its biological plausibility and event-driven characteristic. In particular, recently, many studies on high-performance SNNs comparable to the conventional analog-valued neural networks (ANNs) have been reported by converting weights trained from ANNs into SNNs. However, unlike ANNs, SNNs have an inherent latency that is required to reach the best performance because of differences in operations of neuron. In SNNs, not only spatial integration but also temporal integration exists, and the information is encoded by spike trains rather than values in ANNs. Therefore, it takes time to achieve a steady-state of the performance in SNNs. The latency is worse in deep networks and required to be reduced for the practical applications. In this work, we propose a pre-charged membrane potential (PCMP) for the latency reduction in SNN. A variety of neural network applications (e.g., classification, autoencoder using MNIST and CIFAR-10 datasets) are trained and converted to SNNs to demonstrate the effect of the proposed approach. The latency of SNNs is successfully reduced without accuracy loss. In addition, we propose a delayed evaluation method (DE), by which the errors during the initial transient are discarded. The error spikes occurring in the initial transient is removed by DE, resulting in the further latency reduction. DE can be used in combination with PCMP for further latency reduction. Finally, we also show the advantages of the proposed methods in improving the number of spikes required to reach a steady-state of the performance in SNNs for energy-efficient computing.

Reservoir computing of Spiking Neural Network (SNN) has attracted much attention, which is a recently developed generic cortical model shown to be beneficial for performing computational tasks. Recent studies have demonstrated that the structure of reservoir network (hidden layer) plays an important role on the computational performance of SNN. Here, we propose a novel reservoir network which is multi-clustered generated by a developmental time windows algorithm. Our results show that comparing with random SNN, the model we proposed can significantly reduce the average normalized root mean square error (NRMSE) and standard deviation of errors on the computational performance of Mackey-Glass chaotic time series prediction tasks. Moreover, the influence of cluster size on reservoir computing has been investigated. There exists an optimal value of the cluster size where SNN shows the best computational performance, i.e., NRMSE reaches the minimal value. Our results demonstrate that multi-clustered topology benefits for the computational capability of SNN.

Neuromorphic computer architectures utilize an event-based paradigm to operate in low-power environments and achieve high-throughput with massive parallelism. The integrate-and-fire (IF) neuron model has become a greatest common denominator among many emerging architectures built from materials with neuronlike response properties. There is recent interest in using these architectures as general purpose computing devices, especially for problems that are computationally demanding such as constraint satisfaction, integer factorization, and numerical analysis. However, programming a spiking neural network (SNN) to perform these kinds of tasks remains a difficult problem. There is currently a lack of overarching principles to guide this kind of development, and specific solutions generally remain unportable between different architectures. We identify some of the challenges to using neuromorphic computers for general purpose computing. To address these challenges, we introduce four design principles that aim to facilitate good design on a SNN architecture. We then describe several patterns that solve recurring neuromorphic design challenges, including serial execution on a parallel architecture and the implementation of structured memory in a SNN.

Spiking Neural Networks (SNNs) have emerged as an attractive alternative to traditional deep learning frameworks, since they provide higher computational efficiency in event driven neuromorphic hardware. However, the state-of-the-art (SOTA) SNNs suffer from high inference latency, resulting from inefficient input encoding and training techniques. The most widely used input coding schemes, such as Poisson based rate-coding, do not leverage the temporal learning capabilities of SNNs. This paper presents a training framework for low-latency energy-efficient SNNs that uses a hybrid encoding scheme at the input layer in which the analog pixel values of an image are directly applied during the first timestep and a novel variant of spike temporal coding is used during subsequent timesteps. In particular, neurons in every hidden layer are restricted to fire at most once per image which increases activation sparsity. To train these hybrid-encoded SNNs, we propose a variant of the gradient descent based spike timing dependent backpropagation (STDB) mechanism using a novel cross entropy loss function based on both the output neurons' spike time and membrane potential. The resulting SNNs have reduced latency and high activation sparsity, yielding significant improvements in computational efficiency. In particular, we evaluate our proposed training scheme on image classification tasks from CIFAR-10 and CIFAR-100 datasets on several VGG architectures. We achieve top-l accuracy of 66.46% with 5 timesteps on the CIFAR-100 dataset with ~125x less compute energy than an equivalent standard ANN. Additionally, our proposed SNN performs 5–300 x faster inference compared to other state-of-the-art rate or temporally coded SNN models.

Classification of time series is an essential requirement of various applications that demand continuous monitoring of dynamical systems such as industrial process and health care monitoring. Feature extraction plays a vital role in deciding performance of the time series models. In recent years deep learning techniques have shown an excellent performance to extract highly discriminating features for the classification of the time series. A Convolutional Neural Network (CNN) is a unified framework that performs the feature learning and classification tasks simultaneously. Using the CNN to perform the classification of the multivariate time series is still a challenging task. In this paper, we propose an attention-based multivariate convolutional neural network (AT-MVCNN) that consists of the attention feature-based input tensor scheme to encode informations across the multiple time stamps. The method is capable of learning the temporal characteristics of the multivariate time series. The efficacy of the proposed method is tested on Human Activity Recognition (HAR) and Occupancy Detection datasets. The experiments and results show the proposed method outperforms the other deep learning and traditional machine learning models.

The study is concerned with a concept and a design of granular time series and granular classifiers. In contrast to the plethora of various models of time series, which are predominantly numeric, we propose to effectively exploit the idea of information granules in the description and classification of time series. The numeric (optimization-oriented) and interpretation abilities of granular time series and their classifiers are highlighted and quantified. A general topology of the granular classifier involving a formation of a granular feature space and the usage of the framework of relational structures (relational equations) in the realization of the classifiers is presented. A detailed design process is elaborated on along with a discussion of the pertinent optimization mechanisms. A series of experiments is covered leading to a quantitative assessment of the granular classifiers and their parametric analysis.

With the rapid advancement of deep learning, Spiking Neural Networks (SNNs) have attracted growing interest due to their low power consumption, sensitivity to temporal information, and biological plausibility. However, deploying SNNs in resource-constrained, real-time embedded environments presents significant challenges--chiefly their complex training processes, limited hardware acceleration support, and the difficulty of performing scheduling analysis. This paper presents an integrated modeling and scheduling analysis framework for SNNs based on the MARTE (Modeling and Analysis of Real-Time and Embedded Systems) standard defined by the OMG. Key SNN components--such as neurons, synapses, and spike events--are mapped to schedulable tasks and communication resources within the MARTE profile. Leveraging the Papyrus MARTE tool, we conduct simulation and verification on a heterogeneous embedded platform comprising multi-core ARM and DSP processors. Experimental results show that the proposed framework effectively satisfies end-to-end latency and power constraints, while significantly reducing system integration risks and enhancing design efficiency. Finally, we discuss future research directions, including support for more complex SNN architectures, advanced scheduling strategies, deployment on heterogeneous and distributed platforms, and formal verification for safety-critical applications.

In recent years, spiking neural networks (SNNs), which originated from the theoretical basis of neuroscience, have attracted neuromorphic computing and brain-like computing due to their advantages, such as neural dynamics and coding mechanism, which are similar to biological neurons. SNNs have become one of the mainstream frameworks in the field of brain-like computing. However, most of the Leaky Integrate-and-Fire (LIF) neuron models currently used by SNNs based on direct training of backpropagation (BP) do not consider the changes in the recurrent connections and the dynamic strength of neuron connections over time. This study presented the LIF neuron model with recurrent connections and a method for dynamically changing the presynaptic currents. Recurrent LIF neurons have an additional cyclic connection compared with classic LIF neurons. Their postsynaptic current stimulates a change in membrane potential at the next time point. Their dynamics were more similar to the activities of biological neurons. We also proposed an efficient and flexible BP training method for recurrent LIF neurons. On the basis of the above methods, we proposed the recurrent SNN with dynamic presynaptic currents based on backpropagation (RDS-BP). We test the proposed RDS-BP on three image data sets (MNIST, Fashion-MNIST and CIFAR-10) and two text data sets (IMDB and TREC). The results showed that the performance of RDS-BP not only exceeded the naive SNN models based on BP but also exceeded the SNN methods proposed in previous studies in recent years, which had excellent performance in previous experiments. Our work provides a new LIF neuron model with a recurrent connection and dynamic presynaptic current and a BP training arrangement for the proposed neuron, which could merit developments with neuromorphic and brain-like computing.

An adaptive threshold mechanism for accurate and efficient deep spiking convolutional neural networks

Comparison of spiking neural networks with different topologies based on anti-disturbance ability under external noise

A multifractal formalism relates multiscale quantities to the multifractal spectrum. The multifractal framework provides significant analytical advantages by incorporating a wide range of statistical moment orders (q), thereby enabling a more comprehensive characterization of the intrinsic structural variability embedded in the dataset. The scaling properties of the analyzed rainfall time series was studied using Legendre transformation. This tool is effective for detecting multifractality in the time series of interest and for extracting information on scaling behavior. The obtained parameters may ultimately aid in performing multifractal modeling. The 50-year-long daily rainfall time series shows multifractal properties. The analysis of the generalized Hurst exponent h(q) enabled the classification of time series’ temporal dynamics, distinguishing between persistent, anti-persistent, and uncorrelated behavior. The multifractal analysis proves to be an effective and robust tool to characterize precipitation time series in the context of climate change research. Ultimately, the parameters and features derived from the multifractal spectrum—such as singularity strengths and spectrum width—serve as both quantitative and qualitative metrics for characterizing the spatiotemporal dynamics of rainfall in the semi-arid region of the Central Mexican Plateau.

This paper presents the multiplier-less Modified Leaky Integrate and Fire Neuron Unit (MLIFNU) and proposed a modified Spiking Neural Network (SNN) architecture for Pattern Recognition application. Synthesis results of implementation on Field-Programmable Gate Array are presented. Accordingly, the maximum frequency of the MLIFNU and SNN are 361.991 MHz and 281.442 MHz, respectively.