Tag
This paper presents a Bayesian control framework that integrates spike-based dynamics with probabilistic inference for adaptive control in nonlinear dynamical systems, using a spiking neural network model demonstrated on a benchmark problem.
The paper presents SAGE, a method that adapts surrogate gradients for Spiking Transformers using attention-derived entropy to improve training accuracy, demonstrated on CIFAR-10/100 datasets.
Proposes SG-JEPA, a joint spiking embedding predictive architecture for large-scale dynamic graphs that partitions nodes into context and target sets along the temporal dimension to learn predictive embeddings, achieving competitive performance on node classification while scaling to graphs with 13 million edges and avoiding complex self-supervised mechanisms.
Introduces SpikF-GO, a spiking neural network model for multivariate time series forecasting that combines graph-based inter-variable dependency modeling with spike-driven spectral processing, achieving state-of-the-art results among SNN methods with reduced energy consumption.
Otters++ is a novel optical spiking Transformer that leverages time-to-first-spike coding and physical hardware decay to achieve energy-efficient inference, achieving 84.17% on GLUE while maintaining a clear energy advantage over prior spiking Transformer baselines.
This paper proposes a biologically inspired reinterpretation of surrogate safety measure thresholds using spiking neural networks, aligning with human braking behavior to bridge objective and subjective safety perception in automated driving.
XOResNet introduces OR-ADD shortcut connections and XOR meta-residuals to address spike redundancy and information loss in deep spiking neural networks, achieving state-of-the-art results on Fashion-MNIST, CIFAR-10, CIFAR-100, and miniImageNet.
Introduces Eggroll, a low-rank evolution strategy for gradient-free training of spiking neural networks, reducing memory and time overhead while achieving competitive accuracy on N-MNIST.
This paper proposes a plug-and-play framework that implements spike-friendly approximations for Transformer nonlinearities (e.g., Softmax, SiLU, normalization) via population computation with LIF neurons and lightweight bit-shift scaling, achieving less than 1% accuracy drop on LLMs without fine-tuning.
This paper proposes a federated learning framework for spiking neural networks that addresses the challenge of heterogeneous temporal resolutions across edge devices, enabling collaborative training without sharing raw data while handling temporal mismatches.
Proposes Selective Alignment Knowledge Distillation (SeAl-KD) for Spiking Neural Networks, which selectively aligns class-level and temporal knowledge by equalizing competing logits at erroneous timesteps and reweighting temporal alignment based on confidence and inter-timestep similarity, achieving consistent improvements over existing distillation methods on static and neuromorphic datasets.