Research map / Learning theory and algorithms / Prediction and decision losses
Verified Uncertainty Calibration
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Fit a scaling function, then bin its outputs and average function values within bins instead of averaging labels | Recalibration on CIFAR-10 · CIFAR-10 | marginal calibration error: 35% lower calibration error | Helped |
| Fit a scaling function, then bin its outputs and average function values within bins instead of averaging labels | Recalibration on ImageNet · ImageNet | top-label calibration error: 5x lower top-label calibration error | Helped |
| Use the debiased estimator instead of the plugin estimator | Calibration error estimation · CIFAR-10 | squared calibration error estimation error | Helped |
| Use the debiased estimator instead of the plugin estimator | Calibration error estimation · ImageNet | calibration error estimation accuracy | Helped |
| Increase the number of evaluation bins (ablation) | Binned calibration-error measurement of Platt-scaled models · ImageNet and CIFAR-10 | measured binned calibration error | Mixed |
| Use equal-width binning instead of uniform-mass binning for evaluation (ablation) | Binned calibration-error measurement · ImageNet | ℓ1 calibration error (ECE) | No clear effect |
| Use more bins for ℓ1 calibration-error evaluation (ablation) | Binned calibration-error measurement · CIFAR-10 | ℓ1 calibration error (ECE) | No clear effect |
Closest papers by meaning
- Truthfulness of Calibration Measures (NeurIPS 2024 · similarity 0.86)
- Truthful Calibration Errors for Multi-Class Prediction (NeurIPS 2026 · similarity 0.85)
- Testing Calibration in Nearly-Linear Time (NeurIPS 2024 · similarity 0.85)
- Parametric ρ-Norm Scaling Calibration (AAAI 2025 · similarity 0.85)
- Smooth ECE: Principled Reliability Diagrams via Kernel Smoothing (ICLR 2024 · similarity 0.84)
- Combining Priors with Experience: Confidence Calibration Based on Binomial Process Modeling (AAAI 2025 · similarity 0.84)
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.