Research map / Architectures and efficiency / Spiking networks
Towards Memory- and Time-Efficient Backpropagation for Training Spiking Neural Networks
Qingyan Meng, Mingqing Xiao, Shen Yan, Yisen Wang, Zhouchen Lin, Zhi-Quan Luo
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 |
|---|---|---|---|
| drop temporal gradient components (SLTT) instead of full BPTT | SNN training, ResNet-18 · CIFAR-10 | training memory: 3.00G -> 1.09G | Improved |
| drop temporal gradient components (SLTT) instead of full BPTT | SNN training · CIFAR-10 | accuracy: 94.60% -> 94.59% | No clear effect |
| SLTT instead of BPTT | NF-ResNet-34 · ImageNet | training time: 73.8h -> 66.9h | Improved |
| SLTT instead of BPTT | VGG-11 · DVS-CIFAR10 | accuracy: 73.60% -> 77.30% | Improved |
| SLTT instead of BPTT | VGG-11 · DVS-Gesture | training time: 2.68h -> 2.64h | No clear effect |
| backpropagate at only K=1 random time step (SLTT-1) (ablation) | SLTT, NF-ResNet-34 · ImageNet | training time: 66.90h -> 32.03h | Improved |
| backpropagate at only K=1 random time step (SLTT-1) (ablation) | SLTT, NF-ResNet-34 · ImageNet | accuracy: 66.19% -> 66.17% | No clear effect |
| backpropagate at K=4 steps (SLTT-4) instead of all 20 (ablation) | SLTT, VGG-11 · DVS-Gesture | accuracy: 97.92% -> 97.45% | No clear effect |
| SLTT instead of OTTT | NF-ResNet-34 · ImageNet | training time: 7.52h -> 2.23h | Improved |
| increase time constant tau (larger decay) (ablation) | SLTT · DVS-Gesture and DVS-CIFAR10 | accuracy | Hurt |
| replace batch norm with weight standardization (ablation) | SLTT VGG-11 · DVS-Gesture | accuracy: 97.92% -> 98.50% | Improved |
Closest papers by meaning
- Online Training Through Time for Spiking Neural Networks (NeurIPS 2022 · similarity 0.93)
- Advancing Training Efficiency of Deep Spiking Neural Networks through Rate-based Backpropagation (NeurIPS 2024 · similarity 0.93)
- Enhanced Self-Distillation Framework for Efficient Spiking Neural Network Training (NeurIPS 2025 · similarity 0.91)
- NDOT: Neuronal Dynamics-based Online Training for Spiking Neural Networks (ICML 2024 · similarity 0.91)
- Temporal Flexibility in Spiking Neural Networks: Towards Generalization Across Time Steps and Deployment Friendliness (ICLR 2025 · similarity 0.90)
- Efficient Logit-based Knowledge Distillation of Deep Spiking Neural Networks for Full-Range Timestep Deployment (ICML 2025 · 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.
