Research map / Architectures and efficiency / Spiking networks
Fast-SNN: Fast Spiking Neural Network by Converting Quantized ANN
Yangfan Hu, Qian Zheng, Xudong Jiang, Gang Pan
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 signed IF neuron model (ablation) | native converted SNN (VGG-11 W2A2, T=3) · CIFAR-10 | top-1 accuracy: 84.92 -> 92.80 | Improved |
| add signed IF neuron model (ablation) | native converted SNN (VGG-16 32/3, T=7) · ImageNet | top-1 accuracy: 36.43 -> 72.89 | Improved |
| add layer-wise fine-tuning (ablation) | SNN-beta (VGG-11 W2A2) · CIFAR-10 | top-1 accuracy: 92.80 -> 92.99 | Improved |
| add layer-wise fine-tuning (ablation) | SNN-beta (VGG-16 32/3) · ImageNet | top-1 accuracy: 72.89 -> 72.95 | Improved |
| add layer-wise fine-tuning (ablation) | SNN-beta (2-bit ResNet-56) · CIFAR-10 | top-1 accuracy: 81.25 -> 89.25 | Improved |
| add layer-wise fine-tuning (ablation) | SNN-beta (3-bit ResNet-56) · CIFAR-10 | top-1 accuracy: 88.55 -> 92.17 | Improved |
| add layer-wise fine-tuning (ablation) | SNN-beta (3-bit ResNet-20) · CIFAR-10 | top-1 accuracy: 91.33 -> 92.18 | Improved |
| use learnable firing threshold instead of 99th percentile threshold (ablation) | 2-bit AlexNet SNN, T=3 · CIFAR-10 | top-1 accuracy: 81.66 -> 88.97 | Improved |
| use learnable firing threshold instead of 99th percentile threshold (ablation) | 2-bit VGG-11 SNN, T=3 · CIFAR-10 | top-1 accuracy: 80.48 -> 84.92 | Improved |
| use learnable firing threshold instead of max activation threshold (ablation) | 2-bit VGG-11 SNN, T=3 · CIFAR-10 | top-1 accuracy: 24.04 -> 84.92 | Improved |
| raise latency to 3x7 steps, removing sequential error, with learnable threshold (ablation) | 2-bit AlexNet SNN · CIFAR-10 | top-1 accuracy: 88.97 -> 91.52 | Improved |
| use 2-bit weights in Fast-SNN AlexNet instead of PTL's 4-bit weights | AlexNet · CIFAR-10 | top-1 accuracy: 89.48 -> 91.63 | Improved |
Closest papers by meaning
- Inference-Scale Complexity in ANN-SNN Conversion for High-Performance and Low-Power Applications (CVPR 2025 · similarity 0.94)
- QP-SNN: Quantized and Pruned Spiking Neural Networks (ICLR 2025 · similarity 0.93)
- Are Conventional SNNs Really Efficient? A Perspective from Network Quantization (CVPR 2024 · similarity 0.92)
- Efficient ANN-SNN Conversion with Error Compensation Learning (ICML 2025 · similarity 0.91)
- Faster and Stronger: When ANN-SNN Conversion Meets Parallel Spiking Calculation (ICML 2025 · similarity 0.91)
- Adaptive Calibration: A Unified Conversion Framework of Spiking Neural Networks (AAAI 2025 · similarity 0.91)
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.
