Research map / Learning theory and algorithms / Prediction and decision losses
Predict-then-Calibrate: A New Perspective of Robust Contextual LP
Chunlin Sun, Linyu Liu, Xiaocheng Li
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 PTC-B instead of kNN | robust contextual LP · shortest path, risk 0.8 | average VaR: 2010 -> 1708 | Improved |
| use PTC-B instead of IDCC | robust contextual LP · shortest path, risk 0.8 | average VaR: 2205 -> 1708 | Improved |
| use PTC-B instead of Ellipsoid, which ignores covariates | robust contextual LP · shortest path, risk 0.8 | average VaR: 2535 -> 1708 | Improved |
| use PTC-E instead of PTC-B (ellipsoid vs box uncertainty set) | PTC · shortest path, risk 0.6 | average VaR: 1650 -> 1690 | Hurt |
| use PTC-B instead of DCC | robust contextual LP · fractional knapsack, risk 0.8 | average VaR: -1230 -> -1283 | Improved |
| use PTC-E instead of PTC-B | PTC · fractional knapsack, risk 0.8 | average VaR: -1283 -> -1282 | No clear effect |
| ignore covariates (Ellipsoid) on knapsack | Ellipsoid · fractional knapsack | average VaR: 0 | Hurt |
| use a better prediction model (lower test MSE) | PTC-B · shortest path | average VaR | Improved |
| increase number of training samples T | PTC-B/PTC-E vs DCC/IDCC · shortest path | average VaR | Improved |
| kNN under different risk levels | kNN · knapsack | average coverage: 0.96 | Hurt |
| kNN lacks confidence adjustment on shortest path | kNN · shortest path, risk 0.8 | average coverage: 0.65 | Hurt |
Closest papers by meaning
- Contextual Optimization Under Model Misspecification: A Tractable and Generalizable Approach (ICML 2025 · similarity 0.88)
- Algorithms with Calibrated Machine Learning Predictions (ICML 2025 · similarity 0.87)
- Testing Calibration in Nearly-Linear Time (NeurIPS 2024 · similarity 0.87)
- Leaving the Nest: Going beyond Local Loss Functions for Predict-Then-Optimize (AAAI 2024 · similarity 0.87)
- Smart Surrogate Losses for Contextual Stochastic Linear Optimization with Robust Constraints (NeurIPS 2025 · similarity 0.87)
- Robust Data-driven Prescriptiveness Optimization (ICML 2024 · 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.
