Research map / Learning theory and algorithms / Prediction and decision losses
Conformal Prediction for Distribution-to-Distribution Regression
Trung-Khang Tran, Tuan Hoang, Viet-Hoang Tran, Tan Nguyen
Read the paperCheck your idea against 137,217 papers
AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Extends split conformal prediction to distribution-to-distribution regression, where both inputs and outputs are probability distributions, to provide finite-sample coverage guarantees.
Problem
Constructing finite-sample conformal prediction sets with coverage guarantees when inputs and outputs are both probability distributions (distribution-to-distribution regression).
Approach
Applies split conformal prediction to distribution-valued outputs, extending prior CP work on multilabel, high-dimensional, and functional outputs to the distribution-to-distribution setting. Details beyond the abstract are not stated.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| replace non-adaptive CP with adaptive CP using local calibration | Orthonormal-basis CP for DDR · Truncated Gaussians and Mixture of Betas, NDR/OTR/WR | prediction set size (average q-hat) | Improved |
| replace non-adaptive CP with adaptive CP | Orthonormal-basis CP for DDR · Dark matter 3D and 6D, NDR | prediction set size (efficiency) | Improved |
| replace non-adaptive CP with adaptive CP | Orthonormal-basis CP for DDR · House Price, NDR/OTR/WR | prediction set size | Improved |
| add local calibration for conditional coverage | Adaptive CP · Truncated Gaussians | conditional coverage deviation: deviations bounded by 0.04 | Improved |
| add local calibration for conditional coverage | Adaptive CP · Dark matter 6D | conditional coverage deviation: maximum deviation 0.02 | Improved |
| vary calibration set size (ablation) | Orthonormal-basis CP for DDR · synthetic datasets | empirical coverage | No clear effect |
| raise target coverage level | Both CP variants · synthetic and House Price | prediction set size | Hurt |
Experimental setup
Builds on: split conformal prediction
Closest papers by meaning
- Conformal Prediction via Regression-as-Classification (ICLR 2024 · similarity 0.90)
- A Unified Comparative Study with Generalized Conformity Scores for Multi-Output Conformal Regression (ICML 2025 · similarity 0.90)
- Non-exchangeable Conformal Prediction with Optimal Transport: Tackling Distribution Shift with Unlabeled Data (NeurIPS 2025 · similarity 0.90)
- Optimal transport-based conformal prediction (ICML 2025 · similarity 0.89)
- Kandinsky Conformal Prediction: Beyond Class- and Covariate-Conditional Coverage (ICML 2025 · similarity 0.89)
- Coverage-Guaranteed Prediction Sets for Out-of-Distribution Data (AAAI 2024 · similarity 0.89)
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.
