This work proposes a community-aware sparse topology design framework for graph-based SNNs and challenges the prevailing assumption that network size or sparsity alone is sufficient, demonstrating that how sparse connections are organized - the graph topology - critically influences learning efficiency, accuracy, and energy consumption.
Abstract
Spiking Neural Networks (SNNs) offer a promising pathway toward energy-efficient neuromorphic computing due to their event-driven computation and sparse spike-based communication. However, most existing SNN architectures are derived from dense Artificial Neural Networks (ANNs) and do not explicitly exploit the role of network topology in learning dynamics. In this work, we propose a community-aware sparse topology design framework for graph-based SNNs. Using seven distinct community detection algorithms (KMeans, Spectral Clustering, Fast Greedy, Louvain, Leiden, Infomap, and Small-World), we systematically compare how different modular organizations influence convergence speed, classification accuracy, and energy consumption under strictly controlled conditions (64 neurons, 92% sparsity, ≈ 10 communities, T = 4 time steps). Experimental results on MNIST and CIFAR-10 reveal a dataset-dependent trade-off. On simple, low-noise MNIST, fine-grained methods like Infomap achieve the highest accuracy (99.67%). On the more complex CIFAR-10, coarse and noise-robust methods (Louvain, KMeans, Small-World) perform best (≈ 92.96% accuracy), slightly outperforming fine-grained algorithms (≈ 90.8%). Notably, all community-driven topologies converge dramatically faster than conventional SNNs (27-44 epochs vs. 100-300 epochs). Despite using twice as many neurons as the baseline TANet-Tiny, our sparse modular architectures maintain the same inference energy (≈ 1.2 mJ per sample) thanks to higher sparsity (92% vs. ≈80%) and structured connectivity, halving the energy per neuron. These findings challenge the prevailing assumption that network size or sparsity alone is sufficient, demonstrating that how sparse connections are organized - the graph topology - critically influences learning efficiency, accuracy, and energy consumption. Our framework provides practical guidelines for dataset-aware community detection in neuromorphic system design.
Spiking graph neural networks (SGNNs) have attracted considerable attention due to their high efficiency and low energy consumption in processing graph-structured data. In such networks, the firing threshold of neurons serves as a critical gating mechanism that governs spike sparsity, information flow, and energy consumption. However, existing threshold mechanisms are either fixed global constants or rely solely on coarse-grained statistics such as node degree. These approaches fundamentally fail to capture the rich local structural heterogeneity inherent in graph-structured data. In such data, nodes naturally assume distinct structural roles. Consequently, the resulting spike trains retain insufficient discriminative structural information. This deficiency compromises model capability and robustness against local structural distribution shifts. To address this issue, we propose a structure-adaptive threshold learning framework based on sparse representation. The framework learns a dictionary in which each atom captures a typical local subgraph pattern and is associated with a learnable threshold. Soft weights derived from sparse coding are leveraged to aggregate the atomic thresholds via weighted averaging, yielding a node-specific firing threshold. Furthermore, we design an alternating soft-fusion-hard-grouping training strategy that decouples structure-aware threshold generation from pattern-specific threshold optimization. Experimental results on multiple benchmark semi-supervised node classification tasks demonstrate that the proposed method significantly improves classification accuracy while preserving high spike sparsity. These results validate the effectiveness of the structure-adaptive threshold mechanism for low-power spiking graph learning.
Zehan Li, Yingyi Li, Juntao Zhang et al.· International Journal of Inf...· 0 citations
Efficient communication is central to both biological and artificial intelligence (AI) systems. In biological brains, the challenge of long-range communication across regions is addressed through sparse, spike-based signaling, minimizing energy and latency. Conversely, modern AI workloads are increasingly constrained by bandwidth, leading to bottlenecks that hamper scalability and efficiency. Inspired by the brain’s ability to execute dynamic and complex local computations coupled with sparse inter-neuron communication, we propose heterogeneous neural networks that combine spiking neural networks (SNNs) and artificial neural networks (ANNs) at bandwidth-limited regions, such as chip boundaries, where spike-based communication reduces data transfer overhead. Within each chip, dense ANN computations maintain high throughput, accuracy, and robustness. While SNNs have struggled to algorithmically scale, our approach surmounts this long-standing challenge through algorithm-architecture co-design where learnable sparsity is employed for die-to-die communication by confining spiking layers to specific partitions. This composable design combines high ANN performance with low-bandwidth SNN efficiency. Evaluations on language processing and computer vision exhibit up to 5.3 × energy efficiency gains and 15.2 × latency reductions, surpassing both purely spiking and non-spiking models. As model size grows, improvements scale accordingly. By targeting the inter-chip communication bottleneck with biologically inspired methods, this approach presents a promising path to more efficient AI systems.
Joshua Nardone, Rui-Jie Zhu, Ruhai Lin et al.· International Conference on...· 0 citations
The second-generation neural networks have evolved in recent years, which have become more complex architectures such as spiking neural networks and quantum neural networks. However, the computational resource restriction of neural networks on edge devices is still challenging. The thesis investigates stable learning and compute-resource efficiency on spiking neural networks and hybrid classical-quantum neural networks. Other common qualities like high performance (e.g. high accuracy, high reward), robustness, convergence, predictability, and fast running times were also considered in one or more studies. The contributions of the thesis have several folds. The first study was using audio data; one reason was to verify if a trend called temporal information concentration is present in the spiking neural network. I also gathered other findings, such as dataset complexity impacting Fisher information, related to temporal information dynamics. A second study on spiking neural networks revealed that temporal information concentration was not present in quantization aware training variants, but an increase in Fisher information was found in those variants. The third study on my Multimodal Simplified Spiking Neural Networks explored the effects of audio and image noise. The results show the multimodal model outperformed its unimodal counterparts, but certain configurations of image noises, audio noises, and noise levels performed better than others. In one of the hybrid classical-quantum neural network studies with reinforcement learning, in my quantization-aware training variant, I found higher initial reward growth, longer decreasing in standard deviation and policy entropy, and a few correlations as well related to average reward and policy entropy. In the second study on hybrid classical-quantum neural networks, structured pruning is found to sharpen decisiveness and reveal bad pruning paths, while overparameterization can help exploration. All these studies try to address maintaining or improving stable learning and if the models are computation-resource efficient enough to be realistic.
Spiking neural networks (SNNs) offer a promising pathway to energy-efficient AI and brain-inspired computing. However, their widespread adoption is hindered by a lack of fast, accessible, and versatile simulation frameworks. In this paper, we introduce SuperNeuroMAT, an open-source, scalable, and highly efficient Python-based SNN simulator. We devise a novel matrix-based approach to model the leaky integrate-and-fire (LIF) neuron dynamics and natively support dense and sparse execution modes. This enables fast simulation of approximately 10,000 neurons in dense mode and 100,000 neurons in sparse mode on standard laptops and desktops without requiring specialized hardware. We demonstrate that SuperNeuroMAT consistently outperforms four established SNN simulators---NEST, Brian2, BindsNET, and snnTorch---on two performance metrics (execution speed and peak resident memory) and across various network sizes and connection probabilities. Furthermore, we demonstrate SuperNeuroMAT's applicability across a diverse set of problems. SuperNeuroMAT can efficiently handle conventional machine learning benchmarks such as the Digits and citation network datasets as well as neuromorphic event-based vision tasks such as N-CARS and ASL-DVS. Moreover, it can be extended beyond machine learning workloads and facilitate general-purpose workloads. We validated this by implementing the neuromorphic shortest path algorithm and two arithmetic primitives (addition and multiplication). SuperNeuroMAT can be installed via the Python Package Index (PyPI), thereby lowering the barrier to entry into the field of neuromorphic computing and accelerating the broader development of neuromorphic algorithms.
Prasanna Date, Kevin Zhu, Shruti R. Kulkarni et al.· 0 citations
Spiking neural networks (SNNs), characterized by bio-inspired neuronal dynamics and event-driven communication, have attained significant progress in recent years. Nevertheless, training deep SNNs remains challenging due to spatiotemporal information loss and gradient mismatching. To simultaneously address these issues, we propose a noisy group neuron (NGN) model, which incorporates population-level synchronous resetting and neural stochasticity as fundamental computational mechanisms. We then develop the NGN method as a framework that combines the NGN model with backpropagation learning based on mean-field dynamics. We demonstrate the advantages of the NGN method through theoretical analysis and experimental validation on CIFAR-10, CIFAR-100, Tiny-ImageNet, DVS-Gesture, N-Caltech101, and CIFAR10-DVS. The proposed approach achieves an accuracy of 87.35% on CIFAR10-DVS within 10 inference time steps. These results support NGN as a practical approach to high-performance neuromorphic computing.
Yajie Zhai, Yanmei Kang, Meng Li et al.· 0 citations
Spiking Neural Networks (SNNs) encode information through binary spikes and compute in an event-driven manner, offering an energy-efficient paradigm for machine intelligence. However, high-performance SNNs incur substantial memory and timestep-wise computation costs that hinder deployment on resource-constrained devices. Quantization and pruning provide complementary routes to reducing these costs, yet both make their decisions with local criteria that overlook temporal task feedback in quantization and inter-channel dependencies in pruning. Consequently, optimizing either criterion can still yield suboptimal compression performance. We refer to this discrepancy as criterion-behavior mismatch and propose Behavior-Aligned SNN Compression (BASC), a unified framework with two lightweight modules. For quantization, the scale is applied to synaptic current at every timestep and therefore shifts spike timing. Temporal-Behavior Scale Correction (TSC) makes the scale learnable under a temporal loss, allowing firing behavior to inform scale optimization. For pruning, channel importance depends on how channels jointly drive the membrane potential across the firing threshold. Boundary-Level Inter-Channel Correction (BIC) uses channelwise importance scores for initial selection and inter-channel information to re-evaluate only channels near the pruning threshold. Extensive experiments on static and neuromorphic benchmarks show that lower-bit BASC models match or outperform higher-bit baselines and retain this accuracy advantage after structured pruning, while further reducing model storage and synaptic operations.
Linliang Chen, Yan Zhong, Xin Liu et al.· 0 citations