SPARTA: Advancing Sparse Attention in Spiking Neural Networks via Spike-Timing-Based Prioritization

2025/08/03 by Minsu Jang, Jang, Minsuk, Changick Kim +1 · 1 citation
Engineering · Neuroscience · #Advanced Memory and Neural Computing #Ferroelectric and Negative Capacitance Devices #Neural dynamics and brain function

paper · pdf · doi:10.48550/arxiv.2508.01646

Abstract

Current Spiking Neural Networks (SNNs) underutilize the temporal dynamics inherent in spike-based processing, relying primarily on rate coding while overlooking precise timing information that provides rich computational cues. We propose SPARTA (Spiking Priority Attention with Resource-Adaptive Temporal Allocation), a framework that leverages heterogeneous neuron dynamics and spike-timing information to enable efficient sparse attention. SPARTA prioritizes tokens based on temporal cues, including firing patterns, spike timing, and inter-spike intervals, achieving 65.4% sparsity through competitive gating. By selecting only the most salient tokens, SPARTA reduces attention complexity from O(N²) to O(K²) with k << n, while maintaining high accuracy. Our method achieves state-of-the-art performance on DVS-Gesture (98.78%) and competitive results on CIFAR10-DVS (83.06%) and CIFAR-10 (95.3%), demonstrating that exploiting spike timing dynamics improves both computational efficiency and accuracy.

Citations

• Advancing Spiking Neural Networks towards Multiscale Spatiotemporal Interaction Learning
• QKFormer: Hierarchical Spiking Transformer using Q-K Attention
• Exploring Temporal Information Dynamics in Spiking Neural Networks
• Spikformer: When Spiking Neural Network Meets Transformer
• Training Spiking Neural Networks Using Lessons From Deep Learning
• BSNN: Towards Faster and Better Conversion of Artificial Neural Networks to Spiking Neural Networks with Bistable Neurons
• SpikeMS: Deep Spiking Neural Network for Motion Segmentation
• Progressive Tandem Learning for Pattern Recognition with Deep Spiking Neural Networks
• RMP-SNN: Residual Membrane Potential Neuron for Enabling Deeper High-Accuracy and Low-Latency Spiking Neural Network
• Energy and Policy Considerations for Deep Learning in NLP
• Surrogate Gradient Learning in Spiking Neural Networks
• Surrogate Gradient Learning in Spiking Neural Networks: Bringing the Power of Gradient-Based Optimization to Spiking Neural Networks
• Long short-term memory and learning-to-learn in networks of spiking neurons
• Going Deeper in Spiking Neural Networks: VGG and Residual Architectures
• Deep Residual Learning for Image Recognition
• Why Neurons Have Thousands of Synapses, a Theory of Sequence Memory in Neocortex

Cited by

• DP-MicroAdam: Private and Frugal Algorithm for Training and Fine-tuning

Related

• DTA: Dual Temporal-channel-wise Attention for Spiking Neural Networks
• Spikeformer: A Novel Architecture for Training High-Performance Low-Latency Spiking Neural Network
• Gated Attention Coding for Training High-performance and Efficient Spiking Neural Networks
• Attention Spiking Neural Networks
• STAA-SNN: Spatial-Temporal Attention Aggregator for Spiking Neural Networks