Research map / Learning theory and algorithms / Prediction and decision losses
Conformalized Decision Risk Assessment
Wenbin Zhou, Agni Orfanoudaki, Shixiang Zhu
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 CREDO p-value or e-value radius instead of baselines SA/LR/NN | CREDO · risk estimation (LP, QP, SOCP, IP) | validity: 0.53 -> 1.00 (LP Setting I, SA vs CREDO p) | Improved |
| use Monte Carlo radius (R_inf) instead of p/e-value radius (ablation) | CREDO · risk estimation | validity: 1.00 -> 0.50 (LP Setting I) | Hurt |
| use Monte Carlo radius (R_inf) instead of p/e-value radius (ablation) | CREDO · risk estimation | MAE: 0.27 -> 0.05 (LP Setting I) | Improved |
| use p-value radius instead of e-value radius (ablation) | CREDO · risk estimation | MAE / TPR: 0.31 -> 0.27 (LP Setting I MAE) | Improved |
| use CREDO instead of conformal prediction baseline (CP) | CREDO · QP risk estimation (eps=0.1) | MAE: 0.37 -> 0.16 | Improved |
| use more expressive generative model (3-GMM or VAE instead of 1-GMM) (ablation) | CREDO · decision prescription, LP Setting II | empirical confidence ranking: 3.74 -> 1.05 (sigma=0.1) | Improved |
| use CREDO instead of PTO/RO/SPO+ at very low noise (sigma=0.1) | CREDO · decision prescription, LP Setting I | empirical confidence ranking: 1.00 -> 1.94 | Hurt |
| use CREDO instead of PTO/RO/SPO+ at higher noise (sigma=1) | CREDO · decision prescription, LP Setting I | empirical confidence ranking: 2.76 -> 1.61 | Improved |
| replace oracle/generative model with point prediction in CREDO (ablation) | CREDO · risk estimation | accuracy | Hurt |
| perturb fitted generative model means (ablation) | CREDO · risk estimation | accuracy | Hurt |
| use CREDO alternating optimization instead of GD/BF/RS/RG (ablation) | CREDO · LP Setting I optimization | Err: 0.43 -> 0.00 | Improved |
| use CREDO alternating optimization instead of gradient descent (ablation) | CREDO · QP optimization | Err: 0.50 -> 0.08 | Improved |
Closest papers by meaning
- Decision Theoretic Foundations for Conformal Prediction: Optimal Uncertainty Quantification for Risk-Averse Agents (ICML 2025 · similarity 0.87)
- Utility-Directed Conformal Prediction: A Decision-Aware Framework for Actionable Uncertainty Quantification (ICLR 2025 · similarity 0.87)
- Learning Polyhedral Conformal Sets for Robust Optimization (NeurIPS 2026 · similarity 0.84)
- Conformal Inverse Optimization (NeurIPS 2024 · similarity 0.84)
- Conformal Prediction Sets Improve Human Decision Making (ICML 2024 · similarity 0.84)
- Designing Decision Support Systems using Counterfactual Prediction Sets (ICML 2024 · similarity 0.83)
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.
