Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Bayesian Score Calibration for Approximate Models
Joshua J. Bon, David J. Warne, David J. Nott, Christopher Drovandi
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 |
|---|---|---|---|
| apply Bayesian score calibration (alpha=1) to the approximate posterior | Approx-post · OU process, parameter mu | MSE: 1.54 -> 0.12 | Improved |
| apply Bayesian score calibration (alpha=1) | Approx-post · OU process, parameter mu | achieved coverage of 90% interval (%): 0 -> 82 | Improved |
| apply calibration (alpha=1) to an already accurate approximation | Approx-post · OU process, parameter D | MSE: 4.73 -> 5.13 | No clear effect |
| use full weight clipping alpha=1 instead of no clipping alpha=0 (ablation) | Adjust-post · OU process, parameter mu | achieved coverage of 90% interval (%): 64 -> 82 | Improved |
| use full weight clipping alpha=1 instead of no clipping alpha=0 (ablation) | Adjust-post · OU process, parameter D | achieved coverage of 90% interval (%): 72 -> 83 | Improved |
| apply calibration (alpha=1) | Approx-post · Conjugate Gaussian example | MSE: 0.48 -> 0.14 | Improved |
| apply calibration (alpha=1) | Approx-post · Conjugate Gaussian example | achieved coverage of 90% interval: 0.24 -> 0.98 | Improved |
| clip importance weights at alpha=1 rather than alpha=0 (ablation) | Adjust-post · Conjugate Gaussian example | MSE: 0.21 -> 0.14 | Improved |
| apply calibration (alpha=1) with mean-field variational approximation | Approx-post · Bivariate OU process | mean correlation between rho and D: 0.00 -> 0.31 | Improved |
| use unit weights alpha=1 instead of alpha=0 (ablation) | Adjust-post · Bivariate OU process | mean correlation between rho and D: 0.25 -> 0.31 | Improved |
| clip weights at alpha=0 in higher dimension (ablation) | Adjust-post · Bivariate OU process, parameter D | achieved coverage of 90% interval (%): 84 -> 72 | Hurt |
Closest papers by meaning
- Approximate Bayesian inference from noisy likelihoods with Gaussian process emulated MCMC (JMLR 2024 · similarity 0.89)
- Fast ε-free Inference of Simulation Models with Bayesian Conditional Density Estimation (NeurIPS 2016 · similarity 0.89)
- Calibrated Approximate Bayesian Inference (ICML 2019 · similarity 0.89)
- Learning Likelihood-Free Reference Priors (ICML 2025 · similarity 0.86)
- Bayesian Adaptive Calibration and Optimal Design (NeurIPS 2024 · similarity 0.86)
- Calibration without labels in multiple testing (NeurIPS 2026 · 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.
