Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Sharp Calibrated Gaussian Processes
Alexandre Capone, Sandra Hirche, Geoff Pleiss
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 |
|---|---|---|---|
| Replace variance scaling with optimized calibration hyperparameters | GP regression · Boston | 95% CI width: 0.76 vs 1.4 (RK, RV, RM); NN: 0.73 | Mixed |
| Replace variance scaling with optimized calibration hyperparameters | GP regression · yacht | 95% CI width: 0.76 vs 2.3 (RK, RV, RM); NN: 0.3 | Mixed |
| Replace variance scaling with optimized calibration hyperparameters | GP regression · mpg | 95% CI width: 0.6 vs 1.7 (RK, RM); 1.8 (RV) | Improved |
| Replace variance scaling with optimized calibration hyperparameters | GP regression · wine | 95% CI width: 2.1 vs 3.8 (RK, RV); 3.9 (RM) | Improved |
| Replace variance scaling with optimized calibration hyperparameters | GP regression · concrete | 95% CI width: 0.93 vs 2.5 (RK, RV, RM) | Improved |
| Replace variance scaling with optimized calibration hyperparameters | GP regression · kin8nm | 95% CI width: 0.26 vs 0.47 (RK, RV); 0.48 (RM) | Improved |
| Replace variance scaling with optimized calibration hyperparameters | GP regression · Facebook2 | 95% CI width: 0.6 vs 1.7 (RK, RV, RM) | Improved |
| Use neural-network recalibration instead of Sharp Calibrated GP | GP regression · Boston | ECE: 0.003 -> 0.0056 | Hurt |
| Use neural-network recalibration instead of Sharp Calibrated GP | GP regression · yacht | ECE: 0.0044 -> 0.0081 | Hurt |
| Use neural-network recalibration instead of Sharp Calibrated GP | GP regression · mpg | ECE: 0.0036 -> 0.0053 | Hurt |
| Use neural-network recalibration instead of Sharp Calibrated GP | GP regression · wine | ECE: 0.00047 -> 0.0067 | Hurt |
| Use neural-network recalibration instead of Sharp Calibrated GP | GP regression · concrete | ECE: 0.00071 -> 0.00076 | Hurt |
Closest papers by meaning
- Efficiently sampling functions from Gaussian process posteriors (ICML 2020 · similarity 0.88)
- Goal-Oriented Lower-Tail Calibration of Gaussian Processes for Bayesian Optimization (ICML 2026 · similarity 0.88)
- Practical and Rigorous Uncertainty Bounds for Gaussian Process Regression (AAAI 2021 · similarity 0.87)
- Robust and Conjugate Gaussian Process Regression (ICML 2024 · similarity 0.86)
- Robust and Computation-Aware Gaussian Processes (NeurIPS 2025 · similarity 0.86)
- Bayesian Adaptive Calibration and Optimal Design (NeurIPS 2024 · similarity 0.85)
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.
