Towards efficient and accurate spiking neural networks via adaptive bit allocation.

Predictive coding with spiking neural networks: A survey.

Proactive and privacy-Preserving defense for DNS over HTTPS via federated AI attestation (PAFA-DoH).

HASNN: Hierarchical attention spiking neural network for dynamic graph representation learning

Deep convolutional spiking neural network and block chain based intrusion detection framework for enhancing privacy and security in cloud computing environment

Spiking neural networks (SNNs) have gained significant attention for their potential to enable energy-efficient artificial intelligence (AI). However, effective and efficient training of SNNs remains an unresolved challenge. While backpropagation through time (BPTT) achieves high accuracy, it incurs substantial memory overhead. In contrast, biologically plausible local learning methods are more memory-efficient but struggle to match the accuracy of BPTT. To bridge this gap, we propose spatiotemporal decoupled learning (STDL), a novel training framework that decouples the spatial and temporal dependencies to achieve both high accuracy and training efficiency for SNNs. Specifically, to achieve spatial decoupling, STDL partitions the network into smaller subnetworks, each of which is trained independently using an auxiliary network. To address the decreased synergy among subnetworks resulting from spatial decoupling, STDL constructs each subnetwork's auxiliary network by selecting the largest subset of layers from its subsequent network layers under a memory constraint. Furthermore, STDL decouples dependencies across time steps to enable efficient online learning. Extensive evaluations on seven static and event-based vision datasets demonstrate that STDL consistently outperforms local learning methods and achieves comparable accuracy to the BPTT method with considerably reduced GPU memory cost. Notably, STDL achieves $4\times $ reduced GPU memory than BPTT on the ImageNet dataset. Therefore, this work opens up a promising avenue for memory-efficient SNN training. Code is available at https://github.com/ChenxiangMA/STDL.

Musical mode is a fundamental element of tonal music, structuring pitch organization and shaping tonal relationships. Existing artificial intelligence approaches to symbolic music generation often rely on rigid alignment strategies and simplified tonal representations, limiting their ability to capture the diversity of musical modes, in contrast to the complex perceptual and learning mechanisms observed in human listeners. In this paper, we propose a brain-inspired spiking neural network that integrates biologically grounded mechanisms with symbolic music theory to represent and learn musical modes and keys. The model comprises multiple interacting subsystems inspired by the functional organization of relevant brain regions, and incorporates neural circuit evolution and spike-timing-dependent plasticity to support mode- and key-conditioned music learning and generation. Experimental results show that the synaptic connectivity patterns emerging in the proposed network exhibit strong alignment with the Krumhansl-Schmuckler key profiles, a well-established model of tonal perception in music psychology. Additionally, quantitative evaluations show that the generated musical pieces preserve tonal characteristics while maintaining melodic diversity. By integrating insights from neuroscience, music psychology, and music theory within a spiking neural network framework, this work provides an interpretable and biologically inspired approach to symbolic music learning and generation.

Fast agreement-driven device-calibrated local learning paradigms for spiking neural networks.

Electromagnetic interference (EMI) analysis in high-speed industrial systems is increasingly challenged by multi-gigahertz sampling rates, complex transient behaviors, and stringent real-time constraints. To address these challenges, this paper proposes a pulse-aware generative and analysis framework based on a generative adversarial network (GAN) combined with pulse sparse convolution using leaky integrate-and-fire (LIF) spiking neurons. A multi-scale discriminator and gradient penalty stabilization are employed to improve waveform generation fidelity, achieving a Fréchet distance (FID) of 0.72 and a global difference metric (GDM) of 0.18 ± 0.03 on an industrial-grade Electromagnetic compatibility (EMC) dataset. The proposed framework is further applied to crosstalk prediction, where it reduces pulse-width and phase prediction errors by more than 40% compared with classical numerical solvers such as finite-difference time-domain (FDTD), finite element method (FEM), and method of moments (MoM), and consistently outperforms representative learning-based EMC models. To enable real-time deployment, the pulse sparse convolution architecture is implemented on an field-programmable gate array (FPGA) platform using fixed-point arithmetic, achieving deterministic inference at 5 GS/s with a measured power consumption of 0.71 W. Extensive experiments on traction systems, industrial robots, CNC drives, photovoltaic inverters, and UAV (Unmanned Aerial Vehicle) electronics demonstrate that the proposed approach provides accurate, stable, and energy-efficient EMI analysis suitable for practical industrial EMC applications.

Abstract Anomaly detection offers a promising strategy for discovering new physics at the Large Hadron Collider (LHC). This paper investigates autoencoders (AEs) built using neuromorphic spiking neural networks (SNNs) for this purpose. One key application is at the trigger level, where anomaly detection tools could capture signals that would otherwise be discarded by conventional selection cuts. These systems must operate under strict latency and computational constraints. SNNs are inherently well-suited for low-latency, low-memory, real-time inference, particularly on field-programmable gate arrays. Further gains are expected with the rapid progress in dedicated neuromorphic hardware development. Using the CMS ADC2021 dataset, we design and evaluate a simple SNN AE architecture. Our results show that the SNN AEs are competitive with conventional AEs for LHC anomaly detection across all signal models.

Integrating Neuroscientific Priors into Spiking Neural Networks: ECSNN-SEG for Robust Brain ECS Segmentation from Low-SNR Cryo-electron Microscopy Data

Biological brains mitigate interference by orthogonalizing neural representations of similar memories, thereby preserving stability across tasks in continual learning. However, most existing continual learning approaches for spiking neural networks (SNNs) adopt randomly initialized classifier heads at each step and optimize them with imbalanced data, which often induces representation drift and undermines model stability. In this work, we revisit the role of the classifier head in the continual learning paradigm and propose a pattern separation learning strategy for expandable SNNs in class-incremental learning (CIL). Specifically, we predefine fixed and mutually orthogonal class centers for each class to replace the conventional learnable classifiers, providing stable optimization targets that prevent feature space conflicts and reduce interference between tasks. Combined with dynamically expandable structures that emulate neurogenesis to enhance plasticity, our approach effectively mitigates catastrophic forgetting while maintaining adaptability to novel tasks. Experimental results show that our PS-SNN achieves an average incremental accuracy of 76.42% on the CIFAR100-B0 benchmark over 10 incremental steps. PS-SNN not only surpasses state-of-the-art SNN-based continual learning algorithms but also matches the performance of DNN-based methods, highlighting the potential of integrating biologically inspired pattern separation into neuromorphic computing systems.

A novel rat model harboring two BDNF gene mutations exhibiting autism-like behaviors and cognitive impairments.

Spatially-enhanced Spiking neural network for efficient point cloud analysis.

Efficient speech command recognition leveraging spiking neural networks and progressive time-scaled curriculum distillation.

LAMSNN: Learnable adaptive modulation for artifact suppression in spiking underwater image enhancement networks.

Spiking Depth: Depth estimation from sparse events with spiking neural networks

Spiking neural networks (SNNs), inspired by biological neural mechanisms, represent a promising neuromorphic computing paradigm that offers energy-efficient alternatives to traditional artificial neural networks (ANNs). Despite proven effectiveness, SNN architectures have struggled to achieve competitive performance on large-scale speech processing tasks. Two key challenges hinder progress: 1) the high computational overhead during training caused by multitimestep spike firing and 2) the absence of large-scale SNN architectures tailored to speech processing tasks. To overcome the issues, we introduce the input-aware multilevel spikeformer (IML-Spikeformer), a spiking transformer architecture specifically designed for large-scale speech processing. Central to our design is the input-aware multilevel spike (IMLS) mechanism, which simulates multitimestep spike firing within a single timestep using an adaptive, input-aware thresholding scheme. IML-Spikeformer further integrates a reparameterized spiking self-attention (RepSSA) module with a hierarchical decay mask (HDM), forming the HD-RepSSA module. This module enhances the precision of attention maps and enables modeling of multiscale temporal dependencies in speech signals. Experiments demonstrate that IML-Spikeformer achieves word error rates (WERs) of 6.0% on AiShell-1 and 3.4% on Librispeech-960, comparable to conventional ANN transformers while reducing theoretical inference energy consumption by $4.64\times $ and $4.32\times $ , respectively. IML-Spikeformer marks an advance of scalable SNN architectures for large-scale speech processing in both task performance and energy efficiency. Our source code and model checkpoints are publicly available at github.com/Pooookeman/IML-Spikeformer.

BioMotion-SNN: Spiking neural network modeling for visual motion processing.

Abstract Recurrent spiking neural networks (RSNNs) can be implemented very efficiently in neuromorphic systems. Nevertheless, training of these models with powerful gradient-based learning algorithms is mostly performed on standard digital hardware using Backpropagation through time (BPTT). However, BPTT has substantial limitations. It does not permit online training and its memory consumption scales linearly with the number of computation steps. In contrast, learning methods using forward propagation of gradients operate in an online manner with a memory consumption independent of the number of time steps. These methods enable SNNs to learn from continuous, infinite-length input sequences. In addition, approximate forward propagation algorithms have been developed that can be implemented on neuromorphic hardware. Yet, slow execution speed on conventional hardware as well as inferior performance has hindered their widespread application. In this work, we introduce HYbrid PRopagation (HYPR) that combines the efficiency of parallelization with approximate online forward learning. Our algorithm yields high-throughput online learning through parallelization, paired with constant, i.e., sequence length independent, memory demands. HYPR enables parallelization of parameter update computation over subsequences for RSNNs consisting of almost arbitrary non-linear spiking neuron models. We apply HYPR to networks of spiking neurons with oscillatory subthreshold dynamics. We find that this type of neuron model is particularly well trainable by HYPR, resulting in an unprecedentedly low task performance gap between approximate forward gradient learning and BPTT.