Research map / Point clouds / Point cloud classification
Spiking Discrepancy Transformer for Point Cloud Analysis
Yijie Lu, Zhiyi Pan, Renrui Zhang, Yanhao Jia, Ronggang Wang, Zhaokun Zhou
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 |
|---|---|---|---|
| replace spiking self-attention with SEDA in all stages (ablation) | SDT · ModelNet40 | OA (%): 89.78 -> 92.34 | Improved |
| replace spiking self-attention with SEDA in all stages (ablation) | SDT · ScanObjectNN (PB_T50_RS) | OA (%): 83.20 -> 85.45 | Improved |
| combine SEDA in early stages with SIDA in later stages (ablation) | SDT · ScanObjectNN (PB_T50_RS) | OA (%): 85.45 -> 86.19 | Improved |
| add attention at all (ablation) | SDT · ModelNet40 | OA (%): 85.25 -> 92.46 | Improved |
| add spatially-aware spiking neuron (SASN) (ablation) | SDT · ScanObjectNN (PB_T50_RS) | OA (%): 84.37 -> 86.19 | Improved |
| add spatially-aware spiking neuron (SASN) (ablation) | SDT · ModelNet40 | OA (%): 91.16 -> 92.46 | Improved |
| use SASN instead of IMP neuron (ablation) | SDT · ScanObjectNN (PB_T50_RS) | OA (%): 84.56 -> 86.19 | Improved |
| increase time steps from 1 to 4 (ablation) | SDT · ScanObjectNN (PB_T50_RS) | OA (%): 85.25 -> 86.19 | Improved |
| increase time steps from 4 to 6 (ablation) | SDT · ModelNet40 | OA (%): 92.46 -> 91.93 | Hurt |
| increase depth to 6 stages (ablation) | SDT · ModelNet40 | OA (%): 92.46 -> 92.04 | Hurt |
| use SDAM instead of SSA under spatial transformations | SDT · ModelNet40 | OA drop (%): -3.6 -> -1.3 | Improved |
| use SDT instead of E-3DSNN | SNN · ScanObjectNN (PB_T50_RS) | OA (%): 83.91 -> 86.19 | Improved |
Closest papers by meaning
- Spiking Point Transformer for Point Cloud Classification (AAAI 2025 · similarity 0.90)
- Point-to-Spike Residual Learning for Energy-Efficient 3D Point Cloud Classification (AAAI 2024 · similarity 0.88)
- Noise-Injected Spiking Graph Convolution for Energy-Efficient 3D Point Cloud Denoising (AAAI 2025 · similarity 0.87)
- STEP: A Unified Spiking Transformer Evaluation Platform for Fair and Reproducible Benchmarking (NeurIPS 2025 · similarity 0.87)
- Spiking Transformer: Introducing Accurate Addition-Only Spiking Self-Attention for Transformer (CVPR 2025 · similarity 0.86)
- SpikingResformer: Bridging ResNet and Vision Transformer in Spiking Neural Networks (CVPR 2024 · 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.
