Research map / Time series / Causal and generative time series
Accurate Uncertainties for Deep Learning Using Calibrated Regression
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 isotonic regression recalibration to Bayesian linear regression | Bayesian linear regression · UCI datasets | calibration error: 0.053 -> 0.057 (mpg); 0.175 -> 0.029 (auto); 0.030 -> 0.016 (crime); 0.024 -> 0.006 (kinematics); 0.163 -> 0.052 (stocks | Mixed |
| Apply isotonic regression recalibration to the approximate Bayesian neural network | Approx. Bayesian neural net · UCI datasets | calibration error: 0.102 -> 0.021 (mpg); 0.205 -> 0.017 (auto); 0.070 -> 0.015 (crime); 0.043 -> 0.016 (kinematics); 0.183 -> 0.024 (stocks | Helped |
| Use recalibrated dropout uncertainty estimates | DenseNet · Make3D | calibration error: 5.50E-02 to 2.41E-02 | Helped |
| Use calibrated state-transition uncertainty for planning | Model-based inventory management agent · Inventory management, four stores | cumulative reward: $60,082; $60,894; $61,690 | Helped |
| Replace standard dropout with concrete dropout | Neural network uncertainty models · UCI datasets | calibration error | Mixed |
| Use a deep ensemble instead of the original neural network | Neural network uncertainty models · UCI datasets | calibration error | Mixed |
Closest papers by meaning
- Calibrated Reliable Regression using Maximum Mean Discrepancy (NeurIPS 2020 · similarity 0.90)
- Better Uncertainty Calibration via Proper Scores for Classification and Beyond (NeurIPS 2022 · similarity 0.90)
- Building Calibrated Deep Models via Uncertainty Matching with Auxiliary Interval Predictors (AAAI 2020 · similarity 0.90)
- Uncertainty Quantification with the Empirical Neural Tangent Kernel (NeurIPS 2025 · similarity 0.88)
- Uncertainty Quantification for Deep Regression using Contextualised Normalizing Flows (NeurIPS 2025 · similarity 0.88)
- Reassessing How to Compare and Improve the Calibration of Machine Learning Models (ICLR 2025 · 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.