Research map / Point clouds / Point cloud classification
Point Cloud Processing via Recurrent Set Encoding
Pengxiang Wu, Chao Chen, Jingru Yi, Dimitris Metaxas
This paper tried a common change: Add dropout
Add random point dropout during training · ModelNet40, 1024 points · accuracy: 90.2 -> 91.1 Improved
What 795 accepted papers report for this change
- 549 improved
- 87 no clear effect
- 12 mixed
- 147 hurt
- Don't Drop Dropout: Optimizing Layer Sparsity for Efficient LLM Training and Inference ICML 2026 · apply layer dropout to whole transformer blocks instead of attention and FFN sub-blocks separately · 20 TPP val loss, 271M · val loss: 2.377 -> 2.367Improved
- Quantifying the Uncertainty of Foundation Models with Singular Value Ensembles ICML 2026 · use MC Dropout · vision and SST-2 · accuracy: 86.3 -> 76.6 on Flowers102Hurt
- Motion Planning in Compressed Representation Spaces ICML 2026 · add nested dropout during training · WOMD guided generation · goal FDE: 4.145 -> 1.104Improved
See this change in detailMatched by the wording of each paper's own result, read by AI from the full text. Papers report gains more often than failures, so this shows what gets published, not the chance a change works.
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 the MLP plus global max-pooling encoder with the recurrent set encoder (ablation) | RCNet · ModelNet40 | accuracy: 89.1 -> 91.6 | Improved |
| Ensemble RCNets with beam depths along x, y, and z | RCNet · ModelNet40 | accuracy: 91.6 -> 92.3 | Improved |
| Ensemble RCNets with beam depths along x, y, and z | RCNet · ShapeNet part segmentation | instance average mIoU: 85.3 -> 86.0 | Improved |
| Use 8 × 8 beams rather than 32 × 32 beams (ablation) | RCNet · ModelNet40 | accuracy: 91.6 -> 87.5 | Hurt |
| Use 16 × 16 beams rather than 32 × 32 beams (ablation) | RCNet · ModelNet40 | accuracy: 91.6 -> 90.2 | Hurt |
| Use 64 × 64 beams rather than 32 × 32 beams (ablation) | RCNet · ModelNet40 | accuracy: 91.6 -> 90.9 | Hurt |
| Use 128 × 128 beams rather than 32 × 32 beams (ablation) | RCNet · ModelNet40 | accuracy: 91.6 -> 89.8 | Hurt |
| Add random point dropout during training (ablation) | RCNet · ModelNet40, 1024 points | accuracy: 90.2 -> 91.1 | Improved |
| Add random point dropout during training (ablation) | RCNet · ModelNet40, 512 points | accuracy: 76.2 -> 90.4 | Improved |
| Add random point dropout during training (ablation) | RCNet · ModelNet40, 256 points | accuracy: 38.1 -> 90.2 | Improved |
| Add random point dropout during training (ablation) | RCNet · ModelNet40, 128 points | accuracy: 24.9 -> 87.8 | Improved |
Closest papers by meaning
- Fully-Convolutional Point Networks for Large-Scale Point Clouds (ECCV 2018 · similarity 0.89)
- PointCNN++: Performant Convolution on Native Points (CVPR 2026 · similarity 0.89)
- Cloud Transformers: A Universal Approach to Point Cloud Processing Tasks (ICCV 2021 · similarity 0.89)
- Improved MLP Point Cloud Processing with High-Dimensional Positional Encoding (AAAI 2024 · similarity 0.87)
- LinNet: Linear Network for Efficient Point Cloud Representation Learning (NeurIPS 2024 · similarity 0.87)
- PointRWKV: Efficient RWKV-Like Model for Hierarchical Point Cloud Learning (AAAI 2025 · similarity 0.87)
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.
