Research map / Learning theory and algorithms / Prediction and decision losses
CD-split and HPD-split: Efficient Conformal Regions in High Dimensions
Rafael Izbicki, Gilson Shimizu, Rafael B. Stern
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 |
|---|---|---|---|
| use CD-split+ instead of Reg-split | conformal prediction regions · Photometric redshift (Happy A, faint galaxies) | local coverage: 0.658 -> 0.792 | Improved |
| use HPD-split instead of Reg-split | conformal prediction regions · Photometric redshift (Happy A, faint galaxies) | local coverage: 0.658 -> 0.788 | Improved |
| use CD-split+ instead of Dist-split | conformal prediction regions · Photometric redshift (Happy A, faint galaxies) | local coverage: 0.746 -> 0.792 | Improved |
| use CD-split+ instead of Quantile-split | conformal prediction regions · Photometric redshift (Happy A, faint galaxies) | average size: 0.074 -> 0.066 | Improved |
| use CD-split+ instead of Reg-split | conformal prediction regions · Photometric redshift (Happy A, faint galaxies) | average size: 0.045 -> 0.066 | Hurt |
| use CD-split+ instead of Reg-split | conformal prediction regions · Photometric redshift (Happy A, bright galaxies) | local coverage: 0.807 -> 0.795 | No clear effect |
| increase number of irrelevant features from d=0 to 5000 (ablation) | CD-split+ and HPD-split · Simulated regression (n=1000) | conditional coverage and size | No clear effect |
| use CD-split+ instead of other conformal methods | conformal regression · Simulated regression (d=20) | conditional coverage | Improved |
| use CD-split+ instead of Probability-split | conformal classification · Simulated logistic classification | conditional coverage deviation | Improved |
| use HPD-split instead of CD-split+ | conformal classification · Simulated logistic classification | band size and conditional coverage | Hurt |
Closest papers by meaning
- Neuc-MDS: Non-Euclidean Multidimensional Scaling Through Bilinear Forms (NeurIPS 2024 · similarity 0.79)
- Equivariant Blurring Diffusion for Hierarchical Molecular Conformer Generation (NeurIPS 2024 · similarity 0.79)
- Swallowing the Bitter Pill: Simplified Scalable Conformer Generation (ICML 2024 · similarity 0.79)
- Space Group Conditional Flow Matching (NeurIPS 2026 · similarity 0.77)
- Finsler Multi-Dimensional Scaling: Manifold Learning for Asymmetric Dimensionality Reduction and Embedding (CVPR 2025 · similarity 0.77)
- SILSA: Sliding-Window Slice Latents for Topology-Preserving High-Resolution 3D Generation (NeurIPS 2026 · similarity 0.77)
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.
