Research map / Architectures and efficiency / Spiking networks
Masked Spiking Transformer
Ziqing Wang, Yuetong Fang, Jiahang Cao, Qiang Zhang, Zhongrui Wang, Renjing Xu
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 Random Spike Masking at 75% ratio to the self-attention module (ablation) | MST · CIFAR-10 | accuracy and power: 97.27 -> 97.29 accuracy; power x0.74 | Improved |
| apply Random Spike Masking at 50% ratio (ablation) | Spiking ResNet18 · CIFAR-10 | accuracy: 96.48 -> 92.88 | Hurt |
| apply Random Spike Masking at 75% ratio (ablation) | Spiking VGG16 · CIFAR-10 | accuracy: 95.46 -> 79.09 | Hurt |
| mask the MLP module instead of the self-attention module (ablation) | MST · CIFAR-10/CIFAR-100 | accuracy vs masking ratio | Hurt |
| increase number of masked blocks beyond 2-5 (ablation) | MST · CIFAR-100 | accuracy | Hurt |
| use BatchNorm instead of LayerNorm in ANN-to-SNN conversion (ablation) | MST · Membrane potential distribution | ANN-SNN distribution match | Improved |
| increase time steps from 64 to 256 (ablation) | MST · CIFAR-100 | top-1 accuracy (%): 85.40 -> 86.91 | Improved |
| fine-tune masked student with unmasked teacher (distillation) | MST with RSM · CIFAR | accuracy | Improved |
Closest papers by meaning
- Spiking Transformer: Introducing Accurate Addition-Only Spiking Self-Attention for Transformer (CVPR 2025 · similarity 0.91)
- Spike-driven Transformer V2: Meta Spiking Neural Network Architecture Inspiring the Design of Next-generation Neuromorphic Chips (ICLR 2024 · similarity 0.90)
- TTFSFormer: A TTFS-based Lossless Conversion of Spiking Transformer (ICML 2025 · similarity 0.90)
- SpikedAttention: Training-Free and Fully Spike-Driven Transformer-to-SNN Conversion with Winner-Oriented Spike Shift for Softmax Operation (NeurIPS 2024 · similarity 0.89)
- Faster and Stronger: When ANN-SNN Conversion Meets Parallel Spiking Calculation (ICML 2025 · similarity 0.89)
- Towards Efficient Spiking Transformer: a Token Sparsification Framework for Training and Inference Acceleration (ICML 2024 · 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.
