Research map / Architectures and efficiency / Spiking networks
Stable Spike: Dual Consistency Optimization via Bitwise AND Operations for Spiking Neural Networks
Yongqi Ding, Kunshan Yang, Linze Li, Yiyang Zhang, Mengmeng Jing, Lin Zuo
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add spike map consistency loss L_spike (ablation) | VGG-9 SNN · DVS-Gesture | accuracy (%): 87.15 -> 91.32 | Improved |
| add perturbation consistency loss L_noise (amplitude-aware spike noise) (ablation) | VGG-9 SNN · DVS-Gesture | accuracy (%): 87.15 -> 94.09 | Improved |
| add both consistency losses (ablation) | ResNet-18 SNN · CIFAR10-DVS | accuracy (%): 66.1 -> 70.3 | Improved |
| add both consistency losses (ablation) | QKFormer · CIFAR10-DVS | accuracy (%): 81.2 -> 82.9 | Improved |
| use AND instead of OR to extract stable spikes (ablation) | Stable Spike VGG-9 · CIFAR10-DVS | accuracy (%): 68.9 -> 77.1 | Improved |
| use AND instead of XOR to extract stable spikes (ablation) | Stable Spike VGG-9 · DVS-Gesture | accuracy (%): 89.58 -> 94.44 | Improved |
| use amplitude-aware noise instead of fixed noise probability p=0.5 (ablation) | Stable Spike VGG-9 · DVS-Gesture | accuracy (%): 88.89 -> 94.44 | Improved |
| use discrete spike noise instead of continuous Gaussian noise (std=0.5) (ablation) | Stable Spike VGG-9 · DVS-Gesture | accuracy (%): 91.67 -> 94.44 | Improved |
| use amplitude-aware spike noise instead of adaptive Gaussian noise (ablation) | Stable Spike VGG-9 · DVS-Gesture | accuracy (%): 88.54 -> 94.44 | Improved |
| swap MSE consistency for KL or cosine in L_spike (ablation) | Stable Spike VGG-9 · CIFAR10-DVS | accuracy (%): 77.1 -> 76.2 (KL), 75.6 (Cosine) | No clear effect |
| apply Stable Spike at ultra-low latency T=2 | VGG-9 SNN · DVS-Gesture | accuracy (%): 83.68 -> 92.01 | Improved |
| add AND-based consistency to more layers (Stable Spike Dense) (ablation) | Stable Spike VGG-9 · CIFAR10-DVS | accuracy (%): 77.1 -> 78.8 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- STAA-SNN — STAA-SNN: Spatial-Temporal Attention Aggregator for Spiking Neural Networks (CVPR 2025) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Rethinking Spiking Neural Networks from an Ensemble Learning Perspective (ICLR 2025 · similarity 0.89)
- SPACE: SPike-Aware Consistency Enhancement for Test-Time Adaptation in Spiking Neural Networks (NeurIPS 2025 · similarity 0.88)
- Shrinking Your TimeStep: Towards Low-Latency Neuromorphic Object Recognition with Spiking Neural Networks (AAAI 2024 · similarity 0.88)
- Robust Stable Spiking Neural Networks (ICML 2024 · similarity 0.88)
- Training High Performance Spiking Neural Network by Temporal Model Calibration (ICML 2025 · similarity 0.88)
- Temporal Flexibility in Spiking Neural Networks: Towards Generalization Across Time Steps and Deployment Friendliness (ICLR 2025 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
