Research map / Learning theory and algorithms / Prediction and decision losses
Conformal Prediction for Time-Dependent PDEs
Joshua Stiller, Annika Schneider, Eyke Hüllermeier
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.
Studies conformal prediction for time-dependent PDE surrogates, where temporal dependence can break coverage guarantees and cause systematic undercoverage.
Problem
Conformal prediction's distribution-free coverage guarantees can fail in time-dependent settings common in physics and engineering (e.g., flood forecasting, thermal control), leading to systematic undercoverage of prediction intervals for PDE models.
Approach
Studies the failure of conformal coverage under temporal dependence in PDE forecasting; the abstract does not specify the proposed remedy.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use Gaussian-likelihood-ratio weighted CP (WCP Gauss) instead of naive split CP | conformal prediction for PDE surrogates · synthetic second-order PDEs and three real datasets | coverage (90% target) | Improved |
| estimate only mean and covariance for weights (WCP Gauss) instead of learned weights | weighted conformal prediction · synthetic and real-world PDEs | coverage | Improved |
| use adaptive conformal inference (DtACI) without recalibration | DtACI · recalibration task | coverage | No clear effect |
| refine the prediction grid | weighted and standard CP · drift PDE | coverage and band width | Hurt |
| restrict calibration to a coarse grid or region of interest | CP guarantees · real-world data | coverage guarantee | Improved |
| reduce sample size or raise dimensionality for covariance estimation (ablation) | WCP Gauss · ablation | coverage | Hurt |
| apply WCP Gauss to nonlinear PDEs | WCP Gauss · nonlinear PDEs | coverage | Hurt |
| construct bands from estimated mean and covariance rather than estimating full distribution shift | WCP · real-world data | coverage | Improved |
Closest papers by meaning
- Flow-Based Conformal Predictive Distributions (NeurIPS 2026 · similarity 0.87)
- Calibrated Physics-Informed Uncertainty Quantification (ICML 2025 · similarity 0.87)
- Probabilistic Conformal Prediction with Approximate Conditional Validity (ICLR 2025 · similarity 0.86)
- Conformal Prediction as Bayesian Quadrature (ICML 2025 · similarity 0.86)
- Generative Conformal Prediction with Optimized Coverage Allocation (NeurIPS 2026 · similarity 0.86)
- Compute-optimal data scaling for neural surrogates via multi-fidelity training (NeurIPS 2026 · 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.
