Research map / Architectures and efficiency / Spiking networks
Training Spiking Neural Networks with Local Tandem Learning
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| local layer-wise loss (LTL offline) instead of end-to-end STBP/TET training | SNN VGG-11 · CIFAR-10 | convergence speed | Helped |
| offline LTL vs STBP at short time windows | SNN VGG-11 · CIFAR-10 | accuracy | Helped |
| online (temporally local) instead of offline LTL | LTL VGG-11 LIF Tw=16 · CIFAR-10 | SNN accuracy (%): 93.20 -> 92.11 | Hurt |
| online LTL instead of STBP for memory/time | VGG-11 · CIFAR-10 | GPU memory (MB): 7072 -> 718 | Helped |
| online LTL instead of STBP for training time | VGG-11 · CIFAR-10 | training time per epoch (s): 489.80 -> 103.30 | Helped |
| increase time window from 16 to 32 | Offline LTL LIF VGG-16 · CIFAR-10 | SNN accuracy (%): 93.23 -> 93.92 | Helped |
| increase warm-up steps in online LTL from 2 to 14 (ablation) | Online LTL LIF VGG-16 · CIFAR-10 | SNN accuracy (%): 91.07 -> 93.00 | Helped |
| train layers sequentially instead of in parallel (ablation) | Offline LTL VGG-11 · CIFAR-10 | accuracy (%): 93.20 -> 92.98 | No clear effect |
| LTL on-chip fine-tuning vs HIL under strong device mismatch | VGG-11 SNN · CIFAR-10 mismatch sigma=0.40 | accuracy (%): 82.11 -> 92.26 | Helped |
| LTL vs HIL under neuron silencing p=0.5 | VGG-11 SNN · CIFAR-10 | accuracy (%): 90.54 -> 84.45 | Hurt |
| LTL on-chip calibration under noise (mismatch sigma=0.40) | pre-trained SNN VGG-11 · CIFAR-10 | accuracy (%): 71.20 -> 92.26 | Helped |
Closest papers by meaning
- Enhanced Self-Distillation Framework for Efficient Spiking Neural Network Training (NeurIPS 2025 · similarity 0.90)
- Spike-driven Transformer V2: Meta Spiking Neural Network Architecture Inspiring the Design of Next-generation Neuromorphic Chips (ICLR 2024 · similarity 0.89)
- Enhancing Training of Spiking Neural Network with Stochastic Latency (AAAI 2024 · similarity 0.88)
- Multiplication-Free Parallelizable Spiking Neurons with Efficient Spatio-Temporal Dynamics (NeurIPS 2025 · similarity 0.88)
- Temporal Flexibility in Spiking Neural Networks: Towards Generalization Across Time Steps and Deployment Friendliness (ICLR 2025 · similarity 0.88)
- S$^2$NN: Sub-bit Spiking Neural Networks (NeurIPS 2025 · similarity 0.88)
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.