Research map / Learning theory and algorithms / Matrix and tensor methods
Outlier-Robust Subsampling Techniques for Persistent Homology
Bernadette J. Stolz
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 PH landmarks (I and II) instead of maxmin and random landmarks | PH landmarks · noisy sphere/torus/Klein bottle point clouds, low sampling density | fraction of signal points in landmarks | Improved |
| use PH landmarks instead of maxmin on sphere-Laplace data | PH landmarks · Sphere-Laplace line | fraction of signal points in landmarks | No clear effect |
| exclude outliers and keep only cluster centres as landmarks in k--landmarks (ablation) | k--landmarks · six noisy data sets | fraction of signal points in landmarks | Improved |
| use k--landmarks (outlier-free) instead of random selection at low sampling density | k--landmarks · sphere-cube and Klein bottle | fraction of signal points in landmarks | Mixed |
| restrict PH outlierness to dimension 1 (PH landmarks II, vital landmarks) instead of all dimensions (ablation) | PH landmarks · sphere-line and sphere-Laplace (p=0.6) | fraction of signal points in landmarks | Improved |
| use dimension 1 PH landmarks II instead of PH landmarks I at high signal content (ablation) | PH landmarks · data sets with p > 0.6 | fraction of signal points in landmarks | Improved |
| use PH landmarks I instead of II at low signal content with unstructured noise (ablation) | PH landmarks · unstructured noise, medium/low signal density | fraction of signal points in landmarks | Improved |
| use PH landmarks instead of dense core subsets (K=1) | PH landmarks · most data sets with p=0.6 | fraction of signal points in landmarks | Hurt |
| use PH landmarks II instead of dense core subsets on sphere-Laplace | PH landmarks II · Sphere-Laplace line (p=0.6) | fraction of signal points in landmarks | Improved |
| use smaller delta for PH landmarks (ablation) | PH landmarks · data sets where dense core K=1 does well | fraction of signal points in landmarks | Improved |
| use larger delta for PH landmarks (ablation) | PH landmarks · sphere-line and sphere-Laplace | fraction of signal points in landmarks | Improved |
Closest papers by meaning
- TOPOGRAPH: Topology-Preserving Graph Reduction with Adaptive Structure for Persistent Homology (AAAI 2026 · similarity 0.91)
- Persistent Homology for High-dimensional Data Based on Spectral Methods (NeurIPS 2024 · similarity 0.88)
- Differentiability and Optimization of Multiparameter Persistent Homology (ICML 2024 · similarity 0.87)
- The Flood Complex: Large-Scale Persistent Homology on Millions of Points (NeurIPS 2025 · similarity 0.87)
- Graphcode: Learning from multiparameter persistent homology using graph neural networks (NeurIPS 2024 · similarity 0.86)
- ToMAToMP: Robust and Multi-Parameter Topological Clustering (NeurIPS 2026 · 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.
