Research map / Label-efficient and robust learning / Knowledge distillation
Many Eyes, One Mind: Temporal Multi-Perspective and Progressive Distillation for Spiking Neural Networks
Kai Sun, Peibo Duan, Yongsheng Huang, Nanxu Gong, Levin Kuhlmann
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 TMPD to TWD (ablation) | TWD (ResNet-18, T=4) · CIFAR-10 | top-1 accuracy (%): 95.57 -> 95.98 | Improved |
| add TMPD to TWD (ablation) | TWD (ResNet-18, T=4) · CIFAR-100 | top-1 accuracy (%): 79.10 -> 79.58 | Improved |
| add TPD to TWD (ablation) | TWD (ResNet-18, T=4) · CIFAR-10 | top-1 accuracy (%): 95.57 -> 95.84 | Improved |
| add TPD to TWD (ablation) | TWD (ResNet-18, T=4) · CIFAR-100 | top-1 accuracy (%): 79.10 -> 79.50 | Improved |
| add both TMPD and TPD to TWD (full MEOM) (ablation) | TWD (ResNet-18, T=4) · CIFAR-100 | top-1 accuracy (%): 79.10 -> 79.66 | Improved |
| add TMPD to TWD (ablation) | TWD (ResNet-18) · CIFAR-10 temporal variance Var@4 | Var@4: 0.212 -> 0.133 | Improved |
| add TPD to TWD (ablation) | TWD (ResNet-18) · CIFAR-10 temporal variance Var@4 | Var@4: 0.212 -> 0.199 | Improved |
| use TWD instead of TAD (ablation) | ResNet-18 student · CIFAR-100, T=4 | top-1 accuracy (%): 77.47 -> 79.10 | Improved |
| add TMPD to TWD (ablation) | TWD (ResNet-34 SNN, T=4) · ImageNet | top-1 accuracy (%): 71.04 -> 71.55 | Improved |
| add TPD to TWD (ablation) | TWD (ResNet-34 SNN, T=4) · ImageNet | top-1 accuracy (%): 71.04 -> 71.34 | Improved |
| use TPD over all timesteps (Full) instead of neighbor-step consistency (NSC) (ablation) | TPD (ResNet-18, T=6) · CIFAR-10 | top-1 accuracy (%): 96.19 -> 96.36 | Improved |
| use all timesteps in TPD instead of F-6 A-5 (final step vs five earlier) (ablation) | TPD (ResNet-18, T=6) · CIFAR-10 | top-1 accuracy (%): 95.99 -> 96.36 | 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.
- BKDSNN — BKDSNN: Enhancing the Performance of Learning-based Spiking Neural Networks Training with Blurred Knowledge Distillation (ECCV 2024) · 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
- Efficient Logit-based Knowledge Distillation of Deep Spiking Neural Networks for Full-Range Timestep Deployment (ICML 2025 · similarity 0.93)
- Temporal Separation with Entropy Regularization for Knowledge Distillation in Spiking Neural Networks (CVPR 2025 · similarity 0.92)
- Synergy Between the Strong and the Weak: Spiking Neural Networks are Inherently Self-Distillers (NeurIPS 2025 · similarity 0.91)
- Enhanced Self-Distillation Framework for Efficient Spiking Neural Network Training (NeurIPS 2025 · similarity 0.91)
- Efficient ANN-Guided Distillation: Aligning Rate-based Features of Spiking Neural Networks through Hybrid Block-wise Replacement (CVPR 2025 · similarity 0.90)
- Temporal Flexibility in Spiking Neural Networks: Towards Generalization Across Time Steps and Deployment Friendliness (ICLR 2025 · similarity 0.90)
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.
