Research map / Point clouds / Point cloud classification
Adaptive Topological Feature via Persistent Homology: Filtration Learning for Point Clouds
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Learn an adaptive weighted filtration instead of using Rips filtration | Persistent-homology features for protein classification · Protein classification | accuracy: 79.9 ± 3.0 -> 81.9 ± 2.1 | Helped |
| Learn an adaptive weighted filtration instead of using DTM filtration | Persistent-homology features for protein classification · Protein classification | accuracy: 78.0 ± 1.6 -> 81.9 ± 2.1 | Helped |
| Use persistent-homology features with learned filtration instead of DistMatrixNet features | Protein classification · Protein classification | accuracy: 65.0 ± 12.0 -> 81.9 ± 2.1 | Helped |
| Add learned topological features | DeepSets · ModelNet10 | accuracy: 65.7 ± 1.4 -> 67.5 ± 2.5 | Helped |
| Add learned topological features | PointNet · ModelNet10 | accuracy: 64.3 ± 4.4 -> 68.8 ± 2.0 | Helped |
| Add learned topological features | PointMLP · ModelNet10 | accuracy: 68.8 ± 6.3 -> 60.0 ± 6.3 | Hurt |
| Use learned filtration rather than Rips filtration | DeepSets plus topological features · ModelNet10 | accuracy: 67.0 ± 2.6 -> 67.5 ± 2.5 | No clear effect |
| Use learned filtration rather than Rips filtration | PointNet plus topological features · ModelNet10 | accuracy: 68.4 ± 2.4 -> 68.8 ± 2.0 | Helped |
| Use learned filtration rather than Rips filtration | PointMLP plus topological features · ModelNet10 | accuracy: 57.8 ± 12.4 -> 60.0 ± 6.3 | Helped |
| Use learned filtration rather than DTM filtration | DeepSets plus topological features · ModelNet10 | accuracy: 68.0 ± 2.5 -> 67.5 ± 2.5 | No clear effect |
| Use learned filtration rather than DTM filtration | PointNet plus topological features · ModelNet10 | accuracy: 68.7 ± 2.3 -> 68.8 ± 2.0 | No clear effect |
| Use learned filtration rather than DTM filtration | PointMLP plus topological features · ModelNet10 | accuracy: 57.2 ± 6.8 -> 60.0 ± 6.3 | Helped |
Closest papers by meaning
- Point-Level Topological Representation Learning on Point Clouds (ICML 2025 · similarity 0.91)
- The Flood Complex: Large-Scale Persistent Homology on Millions of Points (NeurIPS 2025 · similarity 0.90)
- Towards Scalable Topological Regularizers (ICLR 2025 · similarity 0.88)
- Scale-Free Image Keypoints Using Differentiable Persistent Homology (ICML 2024 · similarity 0.88)
- Diffeomorphic interpolation for efficient persistence-based topological optimization (NeurIPS 2024 · similarity 0.87)
- Graphcode: Learning from multiparameter persistent homology using graph neural networks (NeurIPS 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.