Spiking Neural Network Architecture Research Articles

Exploring spiking neural networks for deep reinforcement learning in robotic tasks

Spiking Neural Networks (SNNs) stand as the third generation of Artificial Neural Networks (ANNs), mirroring the functionality of the mammalian brain more closely than their predecessors. Their computational units, spiking neurons, characterized by Ordinary Differential Equations (ODEs), allow for dynamic system representation, with spikes serving as the medium for asynchronous communication among neurons. Due to their inherent ability to capture input dynamics, SNNs hold great promise for deep networks in Reinforcement Learning (RL) tasks. Deep RL (DRL), and in particular Proximal Policy Optimization (PPO) has been proven to be valuable for training robots due to the difficulty in creating comprehensive offline datasets that capture all environmental features. DRL combined with SNNs offers a compelling solution for tasks characterized by temporal complexity. In this work, we study the effectiveness of SNNs on DRL tasks leveraging a novel framework we developed for training SNNs with PPO in the Isaac Gym simulator implemented using the skrl library. Thanks to its significantly faster training speed compared to available SNN DRL tools, the framework allowed us to: (i) Perform an effective exploration of SNN configurations for DRL robotic tasks; (ii) Compare SNNs and ANNs for various network configurations such as the number of layers and neurons. Our work demonstrates that in DRL tasks the optimal SNN topology has a lower number of layers than ANN and we highlight how the state-of-art SNN architectures used in complex RL tasks, such as Ant, SNNs have difficulties fully leveraging deeper layers. Finally, we applied the best topology identified thanks to our Isaac Gym-based framework on Ant-v4 benchmark running on MuJoCo simulator, exhibiting a performance improvement by a factor of 4.4× over the state-of-art SNN trained on the same task.

Exploring the synaptic response of reactive Mn electrodes based TiO2 resistive switches

Mn/TiO2/Mn devices, prepared by reactive sputtering and photolithography techniques, were characterized by analyzing their current–voltage (I-V) dependence, non-volatile memory properties, and artificial synapse behavior. The detailed study of its I-V characteristics allowed for highlighting the main conduction mechanisms involved in the electrical transport through the Mn-TiO2 junctions and determining an equivalent circuit model. These results show that the oxidation of metallic Mn electrodes and the application of electrical pulses produce a complex scenario associated with a highly inhomogeneous oxygen vacancy distribution. The resistance hysteresis switching loops were determined, as well as the synaptic-like weight depreciation and potentiation, revealing a linear dependence of the reset voltage as a function of the amplitude of the set voltage and a quasi-linear variation of the conductance with the number of applied pulses. Simulations based on spiking neural network architecture, considering different updates of the synaptic weights, were trained to learn handwriting patterns. Notably, those based on the linear learning rule of the Mn/TiO2/Mn devices outperformed others with increasing non-linear behavior, demonstrating both high recognition and noise tolerance factors, further highlighting the robustness of this approach.

Brain-Inspired Architecture for Spiking Neural Networks.

Spiking neural networks (SNNs), using action potentials (spikes) to represent and transmit information, are more biologically plausible than traditional artificial neural networks. However, most of the existing SNNs require a separate preprocessing step to convert the real-valued input into spikes that are then input to the network for processing. The dissected spike-coding process may result in information loss, leading to degenerated performance. However, the biological neuron system does not perform a separate preprocessing step. Moreover, the nervous system may not have a single pathway with which to respond and process external stimuli but allows multiple circuits to perceive the same stimulus. Inspired by these advantageous aspects of the biological neural system, we propose a self-adaptive encoding spike neural network with parallel architecture. The proposed network integrates the input-encoding process into the spiking neural network architecture via convolutional operations such that the network can accept the real-valued input and automatically transform it into spikes for further processing. Meanwhile, the proposed network contains two identical parallel branches, inspired by the biological nervous system that processes information in both serial and parallel. The experimental results on multiple image classification tasks reveal that the proposed network can obtain competitive performance, suggesting the effectiveness of the proposed architecture.

SNN using color-opponent and attention mechanisms for object recognition

The current spiking neural network (SNN) relies on spike-timing-dependent plasticity (STDP) primarily for shape learning in object recognition tasks, overlooking the equally critical aspect of color information. To address this gap, our study introduces an unsupervised variant of STDP that incorporates principles from color-opponency mechanisms (COM) and classical receptive fields (CRF) found in the biological visual system, facilitating the integration of color information during parameter updates within the SNN architecture. Our approach initially preprocesses images into two distinct feature maps: one for shape and another for color. Then, signals derived from COM and intensity concurrently drive the STDP process, thereby updating parameters associated with both color and shape feature maps. Furthermore, we propose a channel-wise attention mechanism to enhance differentiation among objects sharing similar shapes or colors. Specifically, this mechanism utilizes convolution to generate an output spike-wave, identifying a winner based on earliest spike timing and maximal potential. The winning kernel computes attention, which is then applied via convolution to each input image feature map, generating post-feature maps. A STDP-like normalization rule compares firing times between pre- and post-feature maps, dynamically adjusting channel weights to optimize object recognition during the training phase.We assessed the proposed algorithm using SNN with both single-layer and multi-layer architectures across three datasets. Experimental findings highlight its efficacy and superiority in complex object recognition tasks compared to state-of-the-art (SOTA) algorithms. Notably, our approach achieved a significant 20% performance improvement over the SOTA on the Caltech-101 dataset. Moreover, the algorithm is well-suited for hardware implementation and energy efficiency, leveraging a winner-selection mechanism based on the earliest spike time.

Impact of the La2NiO4+δ Oxygen Content on the Synaptic Properties of the TiN/La2NiO4+δ/Pt Memristive Devices

AbstractThe rapid development of brain‐inspired computing requires new artificial components and architectures for its hardware implementation. In this regard, memristive devices emerged as potential candidates for artificial synapses because of their ability to emulate the plasticity of the biological synapses. In this work, the synaptic behavior of the TiN/La2NiO4+δ/Pt memristive devices based on thermally annealed La2NiO4+δ films is thoroughly investigated. Using electron energy loss spectroscopy (EELS), it is shown that post‐deposition annealing using inert (Ar) or oxidizing (O2) atmospheres affects the interstitial oxygen content (δ) in the La2NiO4+δ films. Electrical characterization shows that both devices exhibit long‐term potentiation/depression (LTP/LTD) and spike‐timing‐dependent plasticity (STDP). At the same time, the Ar annealed TiN/La2NiO4+δ/Pt device demonstrates filamentary‐like behavior, fast switching, and low energy consumption. On the other hand, the O2 annealed TiN/La2NiO4+δ/Pt devices are forming‐free, exhibiting interfacial‐like resistive switching with slower kinetics. Finally, the simulation tools show that spiking neural network (SNN) architectures with weight updates based on the experimental data achieve high inference accuracy in the digit recognition task, which proves the potential of TiN/La2NiO4+δ/Pt devices for artificial synapse applications.

Energy-aware bio-inspired spiking reinforcement learning system architecture for real-time autonomous edge applications.

Mobile, low-cost, and energy-aware operation of Artificial Intelligence (AI) computations in smart circuits and autonomous robots will play an important role in the next industrial leap in intelligent automation and assistive devices. Neuromorphic hardware with spiking neural network (SNN) architecture utilizes insights from biological phenomena to offer encouraging solutions. Previous studies have proposed reinforcement learning (RL) models for SNN responses in the rat hippocampus to an environment where rewards depend on the context. The scale of these models matches the scope and capacity of small embedded systems in the framework of Internet-of-Bodies (IoB), autonomous sensor nodes, and other edge applications. Addressing energy-efficient artificial learning problems in such systems enables smart micro-systems with edge intelligence. A novel bio-inspired RL system architecture is presented in this work, leading to significant energy consumption benefits without foregoing real-time autonomous processing and accuracy requirements of the context-dependent task. The hardware architecture successfully models features analogous to synaptic tagging, changes in the exploration schemes, synapse saturation, and spatially localized task-based activation observed in the brain. The design has been synthesized, simulated, and tested on Intel MAX10 Field-Programmable Gate Array (FPGA). The problem-based bio-inspired approach to SNN edge architectural design results in 25X reduction in average power compared to the state-of-the-art for a test with real-time context learning and 30 trials. Furthermore, 940x lower energy consumption is achieved due to improvement in the execution time.

Accurate and efficient stock market index prediction: an integrated approach based on VMD-SNNs

The stock market index typically mirrors the financial market's performance. Hence, accurate prediction of stock market index trends is essential for investors aiming to mitigate financial risk and enhance future investment returns. Traditional statistical approaches often struggle with the non-linear nature of stock market index data, leading to potential inaccuracies in long-term predictions. To address this issue, we introduce the TCN-LSTM-SNN (TLSNN) model, a hybrid framework that integrates Long Short-Term Memory (LSTM) and Temporal Convolutional Network (TCN) for robust feature extraction, within a highly efficient Spiking Neural Network (SNN) architecture. Additionally, we employ the Subtraction-Average-Based Optimizer (SABO) to refine the Variational Mode Decomposition (VMD) technique, thereby separating the periodic and trend components of stock indices, reducing noise interference, and establishing a decomposition ensemble framework to bolster the model's resilience. The experimental results show that the VMD-TLSNN hybrid model suggested in this study surpasses other individual benchmark models and their hybrid models in prediction accuracy. Additionally, it demonstrates notably lower energy consumption compared to other hybrid models.

Gated parametric neuron for spike-based audio recognition

Spiking neural networks (SNNs) aim to simulate real neural networks in the human brain with biologically plausible neurons. The leaky integrate-and-fire (LIF) neuron is one of the most widely studied SNN architectures. However, it has the vanishing gradient problem when trained with backpropagation. Additionally, its neuronal parameters are often manually specified and fixed, in contrast to the heterogeneity of real neurons in the human brain. This paper proposes a gated parametric neuron (GPN) to process spatio-temporal information effectively with the gating mechanism. Compared with the LIF neuron, the GPN has two distinguishing advantages: (1) it copes well with the vanishing gradients by improving the flow of gradient propagation; and, (2) it learns spatio-temporal heterogeneous neuronal parameters automatically. Additionally, we use the same gate structure to eliminate initial neuronal parameter selection and design a hybrid recurrent neural network-SNN structure. Experiments on two spike-based audio datasets demonstrated that the GPN network outperformed several state-of-the-art SNNs, could mitigate vanishing gradients, and had spatio-temporal heterogeneous parameters. Our work shows the ability of SNNs to handle long-term dependencies and achieve high performance simultaneously.

SNN-BERT: Training-efficient Spiking Neural Networks for energy-efficient BERT

Spiking Neural Networks (SNNs) are naturally suited to process sequence tasks such as NLP with low power, due to its brain-inspired spatio-temporal dynamics and spike-driven nature. Current SNNs employ ”repeat coding” that re-enter all input tokens at each timestep, which fails to fully exploit temporal relationships between the tokens and introduces memory overhead. In this work, we align the number of input tokens with the timestep and refer to this input coding as ”individual coding”. To cope with the increase in training time for individual encoded SNNs due to the dramatic increase in timesteps, we design a Bidirectional Parallel Spiking Neuron (BPSN) with following features: First, BPSN supports spike parallel computing and effectively avoids the issue of uninterrupted firing; Second, BPSN excels in handling adaptive sequence length tasks, which is a capability that existing work does not have; Third, the fusion of bidirectional information enhances the temporal information modeling capabilities of SNNs; To validate the effectiveness of our BPSN, we present the SNN-BERT, a deep direct training SNN architecture based on the BERT model in NLP. Compared to prior repeat 4-timestep coding baseline, our method achieves a 6.46× reduction in energy consumption and a significant 16.1% improvement, raising the performance upper bound of the SNN domain on the GLUE dataset to 74.4%. Additionally, our method achieves 3.5× training acceleration and 3.8× training memory optimization. Compared with artificial neural networks of similar architecture, we obtain comparable performance but up to 22.5× energy efficiency. We would provide the codes.

Direct training high-performance deep spiking neural networks: a review of theories and methods.

Spiking neural networks (SNNs) offer a promising energy-efficient alternative to artificial neural networks (ANNs), in virtue of their high biological plausibility, rich spatial-temporal dynamics, and event-driven computation. The direct training algorithms based on the surrogate gradient method provide sufficient flexibility to design novel SNN architectures and explore the spatial-temporal dynamics of SNNs. According to previous studies, the performance of models is highly dependent on their sizes. Recently, direct training deep SNNs have achieved great progress on both neuromorphic datasets and large-scale static datasets. Notably, transformer-based SNNs show comparable performance with their ANN counterparts. In this paper, we provide a new perspective to summarize the theories and methods for training deep SNNs with high performance in a systematic and comprehensive way, including theory fundamentals, spiking neuron models, advanced SNN models and residual architectures, software frameworks and neuromorphic hardware, applications, and future trends.

Auto-Spikformer: Spikformer architecture search.

The integration of self-attention mechanisms into Spiking Neural Networks (SNNs) has garnered considerable interest in the realm of advanced deep learning, primarily due to their biological properties. Recent advancements in SNN architecture, such as Spikformer, have demonstrated promising outcomes. However, we observe that Spikformer may exhibit excessive energy consumption, potentially attributable to redundant channels and blocks. To mitigate this issue, we propose a one-shot Spiking Transformer Architecture Search method, namely Auto-Spikformer. Auto-Spikformer extends the search space to include both transformer architecture and SNN inner parameters. We train and search the supernet based on weight entanglement, evolutionary search, and the proposed Discrete Spiking Parameters Search (DSPS) methods. Benefiting from these methods, the performance of subnets with weights inherited from the supernet without even retraining is comparable to the original Spikformer. Moreover, we propose a new fitness function aiming to find a Pareto optimal combination balancing energy consumption and accuracy. Our experimental results demonstrate the effectiveness of Auto-Spikformer, which outperforms the original Spikformer and most CNN or ViT models with even fewer parameters and lower energy consumption.

SNNtrainer3D: Training Spiking Neural Networks Using a User-Friendly Application with 3D Architecture Visualization Capabilities

Spiking Neural Networks have gained significant attention due to their potential for energy efficiency and biological plausibility. However, the reduced number of user-friendly tools for designing, training, and visualizing Spiking Neural Networks hinders widespread adoption. This paper presents the SNNtrainer3D v1.0.0, a novel software application that addresses these challenges. The application provides an intuitive interface for designing Spiking Neural Networks architectures, with features such as dynamic architecture editing, allowing users to add, remove, and edit hidden layers in real-time. A key innovation is the integration of Three.js for three-dimensional visualization of the network structure, enabling users to inspect connections and weights and facilitating a deeper understanding of the model’s behavior. The application supports training on the Modified National Institute of Standards and Technology dataset and allows the downloading of trained weights for further use. Moreover, it lays the groundwork for future integration with physical memristor technology, positioning it as a crucial tool for advancing neuromorphic computing research. The advantages of the development process, technology stack, and visualization are discussed. The SNNtrainer3D represents a significant step in making Spiking Neural Networks more accessible, understandable, and easier for Artificial Intelligence researchers and practitioners.

Unleashing the potential of spiking neural networks for epileptic seizure detection: A comprehensive review

Epilepsy is a prevalent neurological condition characterized by repeated seizures resulting from structural or functional changes in the brain. Standard treatment approaches include anti-epileptic medications and surgical interventions. However, for patients with drug-resistant epilepsy, long-term EEG recordings are essential to monitor medication efficacy and make necessary adjustments to minimize seizure frequency. Continuous monitoring of such patients requires compact and comfortable wearable devices. This review explores the potential of Spiking Neural Networks (SNNs) for low-power signal processing in neuromorphic systems, specifically addressing the demand for compact and wearable devices. SNNs offer rapid computation, minimal power consumption, and superior data representation compared to traditional artificial neural networks (ANNs). With biological plausibility and event-driven hardware support, SNNs are well-suited for modeling neural activity in epilepsy. The review delves into existing literature, investigating the potential of SNNs in seizure detection. It covers various aspects, including the neuronal system, neuromorphic computing, SNN architecture, spiking neuron models, spike encoding schemes, and training methodologies. Representative studies utilizing spike-based machine learning for seizure detection are examined. Furthermore, the review addresses gaps, challenges, strengths, weaknesses, and trade-offs in SNN implementations. By spotlighting the capabilities of SNNs, this review aims to stimulate further research and innovative approaches, contributing to the advancement of epilepsy diagnosis and management.

Spintronic Artificial Synapses Using Voltage‐Controlled Multilevel Magnetic States

AbstractNeuromorphic computing offers energy‐efficient computations for large‐scale data processing compared to conventional von Neumann computing. The artificial synapse, a key element for learning and memory operations in neuromorphic computing, requires multi‐state characteristics and the capability to change and store its states. The implementation of hardware‐based artificial synapses using nonvolatile memory provides significant advantages in terms of energy consumption and circuit area compared to their currently employed CMOS‐based counterparts. In this regard, spintronic devices have emerged as a promising candidate due to their desirable properties for artificial synapses, including multilevel formability, non‐volatility, and outstanding writing performance. In this study, spintronic artificial synapses utilizing voltage‐controlled multilevel magnetic states and energy‐ and area‐efficient artificial neural network architectures associated with them are demonstrated. The multilevel states are created by gradually modulating the magnetic easy‐axis orientation from perpendicular to in‐plane and vice versa, which is achieved either by sequentially applying gate voltage pulses or by adjusting the pulse width of the gate voltage. Based on these spintronic artificial synapses, convolutional neural network (CNN) and spiking neural network (SNN) architectures are constructed, demonstrating high recognition accuracy for the MNIST dataset with improved energy efficiency and a reduced circuit area.

A Hybrid Neural Coding Approach for Pattern Recognition With Spiking Neural Networks.

Recently, brain-inspired spiking neural networks (SNNs) have demonstrated promising capabilities in solving pattern recognition tasks. However, these SNNs are grounded on homogeneous neurons that utilize a uniform neural coding for information representation. Given that each neural coding scheme possesses its own merits and drawbacks, these SNNs encounter challenges in achieving optimal performance such as accuracy, response time, efficiency, and robustness, all of which are crucial for practical applications. In this study, we argue that SNN architectures should be holistically designed to incorporate heterogeneous coding schemes. As an initial exploration in this direction, we propose a hybrid neural coding and learning framework, which encompasses a neural coding zoo with diverse neural coding schemes discovered in neuroscience. Additionally, it incorporates a flexible neural coding assignment strategy to accommodate task-specific requirements, along with novel layer-wise learning methods to effectively implement hybrid coding SNNs. We demonstrate the superiority of the proposed framework on image classification and sound localization tasks. Specifically, the proposed hybrid coding SNNs achieve comparable accuracy to state-of-the-art SNNs, while exhibiting significantly reduced inference latency and energy consumption, as well as high noise robustness. This study yields valuable insights into hybrid neural coding designs, paving the way for developing high-performance neuromorphic systems.

Gated Attention Coding for Training High-Performance and Efficient Spiking Neural Networks

Spiking neural networks (SNNs) are emerging as an energy-efficient alternative to traditional artificial neural networks (ANNs) due to their unique spike-based event-driven nature. Coding is crucial in SNNs as it converts external input stimuli into spatio-temporal feature sequences. However, most existing deep SNNs rely on direct coding that generates powerless spike representation and lacks the temporal dynamics inherent in human vision. Hence, we introduce Gated Attention Coding (GAC), a plug-and-play module that leverages the multi-dimensional gated attention unit to efficiently encode inputs into powerful representations before feeding them into the SNN architecture. GAC functions as a preprocessing layer that does not disrupt the spike-driven nature of the SNN, making it amenable to efficient neuromorphic hardware implementation with minimal modifications. Through an observer model theoretical analysis, we demonstrate GAC's attention mechanism improves temporal dynamics and coding efficiency. Experiments on CIFAR10/100 and ImageNet datasets demonstrate that GAC achieves state-of-the-art accuracy with remarkable efficiency. Notably, we improve top-1 accuracy by 3.10% on CIFAR100 with only 6-time steps and 1.07% on ImageNet while reducing energy usage to 66.9% of the previous works. To our best knowledge, it is the first time to explore the attention-based dynamic coding scheme in deep SNNs, with exceptional effectiveness and efficiency on large-scale datasets. Code is available at https://github.com/bollossom/GAC.

A Deep Convolutional Spiking Neural Network for embedded applications

Deep neural networks (DNNs) have received a great deal of interest in solving everyday tasks in recent years. However, their computational and energy costs limit their use on mobile and edge devices. The neuromorphic computing approach called spiking neural networks (SNNs) represents a potential solution for bridging the gap between performance and computational expense. Despite the potential benefits of energy efficiency, the current SNNs are being used with datasets such as MNIST, Fashion-MNIST, and CIFAR10, limiting their applications compared to DNNs. Therefore, the applicability of SNNs to real-world applications, such as scene classification and forecasting epileptic seizures, must be demonstrated yet. This paper develops a deep convolutional spiking neural network (DCSNN) for embedded applications. We explore a convolutional architecture, Visual Geometry Group (VGG16), to implement deeper SNNs. To train a spiking model, we convert the pre-trained VGG16 into corresponding spiking equivalents with nearly comparable performance to the original one. The trained weights of VGG16 were then transferred to the equivalent SNN architecture while performing a proper weight–threshold balancing. The model is evaluated in two case studies: land use and land cover classification, and epileptic seizure detection. Experimental results show a classification accuracy of 94.88%, and seizure detection specificity of 99.45% and a sensitivity of 95.06%. It is confirmed that conversion-based training SNNs are promising, and the benefits of DNNs, such as solving complex and real-world problems, become available to SNNs.

Comparison of Bagging and Sparcity Methods for Connectivity Reduction in Spiking Neural Networks with Memristive Plasticity

Developing a spiking neural network architecture that could prospectively be trained on energy-efficient neuromorphic hardware to solve various data analysis tasks requires satisfying the limitations of prospective analog or digital hardware, i.e., local learning and limited numbers of connections, respectively. In this work, we compare two methods of connectivity reduction that are applicable to spiking networks with local plasticity; instead of a large fully-connected network (which is used as the baseline for comparison), we employ either an ensemble of independent small networks or a network with probabilistic sparse connectivity. We evaluate both of these methods with a three-layer spiking neural network, which are applied to handwritten and spoken digit classification tasks using two memristive plasticity models and the classical spike time-dependent plasticity (STDP) rule. Both methods achieve an F1-score of 0.93–0.95 on the handwritten digits recognition task and 0.85–0.93 on the spoken digits recognition task. Applying a combination of both methods made it possible to obtain highly accurate models while reducing the number of connections by more than three times compared to the basic model.

Benchmarking Artificial Neural Network Architectures for High-Performance Spiking Neural Networks.

Organizations managing high-performance computing systems face a multitude of challenges, including overarching concerns such as overall energy consumption, microprocessor clock frequency limitations, and the escalating costs associated with chip production. Evidently, processor speeds have plateaued over the last decade, persisting within the range of 2 GHz to 5 GHz. Scholars assert that brain-inspired computing holds substantial promise for mitigating these challenges. The spiking neural network (SNN) particularly stands out for its commendable power efficiency when juxtaposed with conventional design paradigms. Nevertheless, our scrutiny has brought to light several pivotal challenges impeding the seamless implementation of large-scale neural networks (NNs) on silicon. These challenges encompass the absence of automated tools, the need for multifaceted domain expertise, and the inadequacy of existing algorithms to efficiently partition and place extensive SNN computations onto hardware infrastructure. In this paper, we posit the development of an automated tool flow capable of transmuting any NN into an SNN. This undertaking involves the creation of a novel graph-partitioning algorithm designed to strategically place SNNs on a network-on-chip (NoC), thereby paving the way for future energy-efficient and high-performance computing paradigms. The presented methodology showcases its effectiveness by successfully transforming ANN architectures into SNNs with a marginal average error penalty of merely 2.65%. The proposed graph-partitioning algorithm enables a 14.22% decrease in inter-synaptic communication and an 87.58% reduction in intra-synaptic communication, on average, underscoring the effectiveness of the proposed algorithm in optimizing NN communication pathways. Compared to a baseline graph-partitioning algorithm, the proposed approach exhibits an average decrease of 79.74% in latency and a 14.67% reduction in energy consumption. Using existing NoC tools, the energy-latency product of SNN architectures is, on average, 82.71% lower than that of the baseline architectures.

Efficient spiking neural network design via neural architecture search

Spiking neural networks (SNNs) are brain-inspired models that utilize discrete and sparse spikes to transmit information, thus having the property of energy efficiency. Recent advances in learning algorithms have greatly improved SNN performance due to the automation of feature engineering. While the choice of neural architecture plays a significant role in deep learning, the current SNN architectures are mainly designed manually, which is a time-consuming and error-prone process. In this paper, we propose a spiking neural architecture search (NAS) method that can automatically find efficient SNNs. To tackle the challenge of long search time faced by SNNs when utilizing NAS, the proposed NAS encodes candidate architectures in a branchless spiking supernet which significantly reduces the computation requirements in the search process. Considering that real-world tasks prefer efficient networks with optimal accuracy under a limited computational budget, we propose a Synaptic Operation (SynOps)-aware optimization to automatically find the computationally efficient subspace of the supernet. Experimental results show that, in less search time, our proposed NAS can find SNNs with higher accuracy and lower computational cost than state-of-the-art SNNs. We also conduct experiments to validate the search process and the trade-off between accuracy and computational cost.