Research map / Learning theory and algorithms / Prediction and decision losses
Better Uncertainty Calibration via Proper Scores for Classification and Beyond
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 |
|---|---|---|---|
| apply Platt scaling to predicted variance, guided by DSS proper calibration error | mixture density network · Residential Building variance regression | avg predicted variance vs MSE: 0.83 -> 11.04 (MSE 10.48) | Helped |
| recalibrate predicted variance | mixture density network · Residential Building, iteration 2000 | SE/variance ratio: 3.51 -> 0.79 | Helped |
| use root calibration upper bound (RBS) instead of ECE/binned/KDE estimators to estimate recalibration improvement | calibration estimation · CIFAR10/CIFAR100/ImageNet recalibration (TS/ETS/DIAG) | robustness of estimated improvement to test set size | Helped |
| use DSS-induced proper calibration error instead of squared KCE to monitor variance recalibration | variance regression · Residential Building | tracking variance recalibration | Helped |
Closest papers by meaning
- Uncertainty Weighted Gradients for Model Calibration (CVPR 2025 · similarity 0.90)
- Towards Certification of Uncertainty Calibration under Adversarial Attacks (ICLR 2025 · similarity 0.90)
- A Benchmark Study on Calibration (ICLR 2024 · similarity 0.89)
- Reassessing How to Compare and Improve the Calibration of Machine Learning Models (ICLR 2025 · similarity 0.89)
- SURE: SUrvey REcipes for building reliable and robust deep networks (CVPR 2024 · similarity 0.88)
- Truthful Calibration Errors for Multi-Class Prediction (NeurIPS 2026 · similarity 0.88)
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.