Research map / Architectures and efficiency / Spiking networks
Neuromorphic Data Augmentation for Training Spiking Neural Networks
Yuhang Li, Youngeun Kim, Hyoungseob Park, Tamar Geller, Priyadarshini Panda
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 geometric NDA (M1N2) to tdBN training | tdBN ResNet-19 · CIFAR10-DVS | top-1 accuracy: 67.9 -> 78.0 | Improved |
| add geometric NDA to tdBN training | tdBN ResNet-19 · N-Caltech 101 | top-1 accuracy: 62.8 -> 78.6 | Improved |
| add NDA to tdBN training | tdBN ResNet-19 · N-Cars | top-1 accuracy: 82.4 -> 87.2 | Improved |
| add NDA to tdBN training | tdBN VGG-11 (128x128) · CIFAR10-DVS | top-1 accuracy: 76.3 -> 81.7 | Improved |
| use geometric instead of photometric/color augmentation (ablation) | NDA ResNet-19 · CIFAR10-DVS | top-1 accuracy: 62.8 -> 78.0 | Improved |
| use geometric instead of photometric/color augmentation (ablation) | NDA ResNet-19 · N-Caltech101 | top-1 accuracy: 64.0 -> 78.6 | Improved |
| increase augmentation intensity/number from M1N2 to M3N3 (ablation) | NDA ResNet-19 · CIFAR10-DVS | top-1 accuracy: 78.0 -> 71.4 | Hurt |
| increase augmentation intensity/number from M1N2 to M3N3 (ablation) | NDA ResNet-19 · N-Caltech101 | top-1 accuracy: 78.6 -> 65.1 | Hurt |
| lower intensity from M1N2 to M1N1 (ablation) | NDA ResNet-19 · CIFAR10-DVS | top-1 accuracy: 78.0 -> 73.4 | Hurt |
| add NDA to training | tdBN ResNet-19 · CIFAR10-DVS weight noise N(0,0.01) | accuracy drop: 19.4% drop -> 1.5% drop | Improved |
| add NDA to training | tdBN ResNet-19 · CIFAR10-DVS weight noise N(0,0.02) | accuracy drop: 43.4% drop -> 11.3% drop | Improved |
| supervised pre-training on N-Caltech101 before finetuning | VGG-11 · CIFAR10-DVS | top-1 accuracy: 81.7 -> 80.9 | No clear effect |
Closest papers by meaning
- An Efficient Knowledge Transfer Strategy for Spiking Neural Networks from Static to Event Domain (AAAI 2024 · similarity 0.90)
- Shrinking Your TimeStep: Towards Low-Latency Neuromorphic Object Recognition with Spiking Neural Networks (AAAI 2024 · similarity 0.87)
- NDOT: Neuronal Dynamics-based Online Training for Spiking Neural Networks (ICML 2024 · similarity 0.86)
- EventZoom: A Progressive Approach to Event-Based Data Augmentation for Enhanced Neuromorphic Vision (AAAI 2025 · similarity 0.86)
- EnOF-SNN: Training Accurate Spiking Neural Networks via Enhancing the Output Feature (NeurIPS 2024 · similarity 0.86)
- Self-cross Feature based Spiking Neural Networks for Efficient Few-shot Learning (ICML 2025 · similarity 0.86)
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.
