Research map / Architectures and efficiency / Spiking networks
Membrane Potential Batch Normalization for Spiking Neural Networks
Yufei Guo, Yuhan Zhang, Yuanpei Chen, Weihang Peng, Xiaode Liu, Liwen Zhang, Xuhui Huang, Zhe Ma
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 MPBN to spiking ResNet20 (T=1) (ablation) | spiking ResNet20 · CIFAR-10 | top-1 accuracy (%): 90.40% -> 92.22% | Improved |
| add MPBN to spiking ResNet20 (T=2) (ablation) | spiking ResNet20 · CIFAR-10 | top-1 accuracy (%): 92.80% -> 93.54% | Improved |
| add MPBN to spiking ResNet20 (T=4) (ablation) | spiking ResNet20 · CIFAR-10 | top-1 accuracy (%): 93.85% -> 94.28% | Improved |
| add MPBN to spiking ResNet20 (T=1) (ablation) | spiking ResNet20 · CIFAR-100 | top-1 accuracy (%): 67.94% -> 68.36% | Improved |
| add MPBN to spiking ResNet20 (T=4) (ablation) | spiking ResNet20 · CIFAR-100 | top-1 accuracy (%): 71.77% -> 72.30% | Improved |
| add MPBN: gain shrinks as time steps increase (ablation) | spiking ResNet20 · CIFAR-10 | top-1 accuracy (%): 90.40 -> 92.22 (T=1) vs 93.85 -> 94.28 (T=4) | Improved |
| use element-wise instead of channel-wise MPBN (ablation) | MPBN ResNet20 (T=4) · CIFAR-10 | top-1 accuracy (%): 94.28% -> 94.42% | Improved |
| use element-wise instead of channel-wise MPBN (ablation) | MPBN ResNet20 (T=4) · CIFAR-100 | top-1 accuracy (%): 72.30% -> 72.49% | Improved |
| add MPBN: flatter loss landscape and faster convergence (ablation) | spiking ResNet20 (T=2) · CIFAR-10/100 | convergence speed / loss landscape | Improved |
| MPBN with ResNet34 vs SEW ResNet34 | MPBN ResNet34 (T=4) · ImageNet | top-1 accuracy (%): 64.71% vs 67.04% (SEW ResNet34) | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- Real Spike — Real Spike: Learning Real-Valued Spikes for Spiking Neural Networks (ECCV 2022) · better on 4 of 4 shared results
- GLIF — GLIF: A Unified Gated Leaky Integrate-and-Fire Neuron for Spiking Neural Networks (NeurIPS 2022) · better on 1 of 1 shared result
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
- Temporal Effective Batch Normalization in Spiking Neural Networks (NeurIPS 2022 · similarity 0.91)
- TAB: Temporal Accumulated Batch Normalization in Spiking Neural Networks (ICLR 2024 · similarity 0.90)
- Going Deeper With Directly-Trained Larger Spiking Neural Networks (AAAI 2021 · similarity 0.89)
- Training Deep Spiking Neural Networks without Normalization (ICML 2026 · similarity 0.89)
- Rethinking the Membrane Dynamics and Optimization Objectives of Spiking Neural Networks (NeurIPS 2024 · similarity 0.89)
- Optimized Potential Initialization for Low-Latency Spiking Neural Networks (AAAI 2022 · similarity 0.89)
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.
