Research map / Point clouds / Point cloud classification
GD-MAE: Generative Decoder for MAE Pre-Training on LiDAR Point Clouds
Honghui Yang, Tong He, Jiaheng Liu, Hua Chen, Boxi Wu, Binbin Lin, Xiaofei He, Wanli Ouyang
This paper tried a common change: Train longer
Increase pre-training epochs from 30 to 120 · Waymo validation · Vehicle AP: 66.54 -> 66.89 Improved
What 387 accepted papers report for this change
- 211 improved
- 62 no clear effect
- 30 mixed
- 84 hurt
- Lightweight Federated Incremental Learning via Decoupled Replay ICML 2026 · increase local training epochs from 5 to 40 · CIFAR10 (α=1.0) · test accuracyImproved
- When Do Hallucinations Arise? A Graph Perspective on the Evolution of Path Reuse and Path Compression ICML 2026 · train longer on the full path search space (overfitting stage) · SBM graph, 10 communities, pin=0.6 · accuracy: 1.0 -> 0.98Hurt
- Causes and Consequences of Representational Similarity in Machine Learning Models ICML 2026 · train longer on LLM continued pretraining to test insufficient adaptation · perplexity across overlap settings · perplexityImproved
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 |
|---|---|---|---|
| Use GD-MAE pre-training (ablation) | SPT detector · ONCE validation | mAP: 62.62 -> 64.92 (+2.30) | Improved |
| Use GD-MAE pre-training (ablation) | Waymo detector · Waymo validation, 5% labeled data | Vehicle mAPH/L2: 59.97 -> 63.86; +3.89 | Improved |
| Add a pyramid architecture (ablation) | SPT detector · Waymo validation | Vehicle APH/L2: 62.10/61.61 -> 64.20/63.70 | Improved |
| Add a shortcut connection (ablation) | Pyramid SPT · Waymo validation | Vehicle APH/L2: 64.20/63.70 -> 66.02/65.55 | Improved |
| Add GD-MAE pre-training (ablation) | Pyramid SPT with shortcut · Waymo validation | Vehicle APH/L2: 66.02/65.55 -> 67.00/66.54 | Improved |
| Use patch-wise rather than block-wise masking (ablation) | GD-MAE · Waymo validation, 5% labeled data | Vehicle AP: 61.60 -> 62.65 | Improved |
| Use point-wise rather than patch-wise masking (ablation) | GD-MAE · Waymo validation, 5% labeled data | Vehicle AP: 62.65 -> 62.41 | Hurt |
| Replace the transformer decoder with the generative decoder (ablation) | MAE pre-training · Waymo validation | Vehicle APH/L2: 66.24 -> 66.54 | Improved |
| Replace the transformer decoder with the generative decoder (ablation) | MAE pre-training · Waymo decoder runtime | runtime: 27.1ms -> 3.2ms | Improved |
| Increase the pre-training data from 20% to 100% (ablation) | GD-MAE · Waymo validation | mAPH/L2: 67.14 -> 67.64 | Improved |
| Increase the labeled-data fraction from 20% to 100% (ablation) | GD-MAE fine-tuning · Waymo validation | Vehicle mAPH/L2: 67.14 -> 67.64 | No clear effect |
| Increase pre-training epochs from 30 to 120 (ablation) | GD-MAE · Waymo validation | Vehicle AP: 66.54 -> 66.89 | Improved |
Later papers that compared with this method
2 later accepted papers report GD-MAE as a baseline in their results tables; 2 of them report a better number on most of the results they share with it.
- T-MAE: Temporal Masked Autoencoders for Point Cloud Representation Learning (ECCV 2024 · better on 3 of 3 shared results)
- BEV-MAE: Bird’s Eye View Masked Autoencoders for Point Cloud Pre-training in Autonomous Driving Scenarios (AAAI 2024 · better on 2 of 2 shared results)
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Multi-Scale Neighborhood Occupancy Masked Autoencoder for Self-Supervised Learning in LiDAR Point Clouds (CVPR 2025 · similarity 0.90)
- 3D Feature Prediction for Masked-AutoEncoder-Based Point Cloud Pretraining (ICLR 2024 · similarity 0.89)
- PCP-MAE: Learning to Predict Centers for Point Masked Autoencoders (NeurIPS 2024 · similarity 0.88)
- GBlobs: Explicit Local Structure via Gaussian Blobs for Improved Cross-Domain LiDAR-based 3D Object Detection (CVPR 2025 · similarity 0.88)
- BEV-MAE: Bird’s Eye View Masked Autoencoders for Point Cloud Pre-training in Autonomous Driving Scenarios (AAAI 2024 · similarity 0.88)
- RI-MAE: Rotation-Invariant Masked AutoEncoders for Self-Supervised Point Cloud Representation Learning (AAAI 2025 · 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.
