Research map / Point clouds / Shape and surface completion
Deep Loss Convexification for Learning Iterative Models
Ziming Zhang, Youjia Shao, Yiqing Zhang, Fangzhou Lin, Haichong K. Zhang, Elke Angelika Rundensteiner
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 DLC to LSTM training | LSTM · Pixel-MNIST | accuracy: 98.2 -> 99.50 | Improved |
| add DLC to DCP | DCP · ModelNet40 unseen classes (w/o ICP) | MSE(Euler): 6.8747 -> 1.0836 | Improved |
| add DLC to PRNet | PRNet · ModelNet40 unseen classes (w/o ICP) | MSE(Euler): 20.3915 -> 11.3448 | Improved |
| add DLC to DCP | DCP · ModelNet40 unseen classes (w/o ICP) | MSE(T): 1.79e-5 -> 2.55e-5 | Hurt |
| add DLC to PRNet | PRNet · 3DMatch unseen (w/o ICP) | MSE(Euler): 1104.9437 -> 294.2320 | Improved |
| add DLC to PRNet | PRNet · 3DMatch unseen (with ICP) | MSE(T): 0.2076 -> 0.0021 | Improved |
| add ICP refinement to DLC+DCP | DLC+DCP · ModelNet40 unseen classes | MSE(Euler): 1.0836 -> 0.1705 | Improved |
| add DLC to PRNet pretrained on ModelNet40 | PRNet · ShapeNetCore unseen class 14 (w/o ICP) | MSE(Euler): 2.0465 -> 1.0666 | Improved |
| add DLC to PRNet pretrained on ModelNet40 | PRNet · ShapeNetCore all classes (w/o ICP) | MSE(Euler): 13.6229 -> 15.3662 | Hurt |
| add hinge-loss constraints (star-convexity) to LK loss (ablation) | DeepLK · Google Earth multimodal alignment | overfitting / pixel error | Improved |
| add all three constraints versus subsets (ablation) | DLC-DeepLK · Google Earth multimodal alignment | success rate | Improved |
| increase rho from small to large (ablation) | DLC-DeepLK · Google Earth multimodal alignment | pixel error | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- PRNet — PRNet: Self-Supervised Learning for Partial-to-Partial Registration (NeurIPS 2019) · better on 12 of 12 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
- Deep Closest Point: Learning Representations for Point Cloud Registration (ICCV 2019 · similarity 0.88)
- PointCNN++: Performant Convolution on Native Points (CVPR 2026 · similarity 0.87)
- RPM-Net: Robust Point Matching Using Learned Features (CVPR 2020 · similarity 0.87)
- APML: Adaptive Probabilistic Matching Loss for Robust 3D Point Cloud Reconstruction (NeurIPS 2025 · similarity 0.86)
- Where Precision Meets Efficiency: Transformation Diffusion Model for Point Cloud Registration (AAAI 2025 · similarity 0.85)
- Inferring Neural Signed Distance Functions by Overfitting on Single Noisy Point Clouds through Finetuning Data-Driven based Priors (NeurIPS 2024 · similarity 0.85)
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.
