Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
High-Quality Prediction Intervals for Deep Learning: A Distribution-Free, Ensembled Approach
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 |
|---|---|---|---|
| Use QD-Ens instead of MVE-Ens | Ensembled neural-network prediction intervals · Ten benchmark regression datasets | MPIW: reducing average PI width by around 10%; MPIW was on average 11.6% narrower | Helped |
| Use QD-Ens instead of MVE-Ens | Ensembled neural-network prediction intervals · Kin8nm | MPIW: 1.14 ± 0.01 -> 1.25 ± 0.01 | Hurt |
| Train QD with gradient descent rather than PSO (ablation) | QD prediction intervals · Synthetic regression data | MPIW | Helped |
| Use the distribution-free QD loss rather than Gaussian-assumption MVE (ablation) | Neural-network prediction intervals · Synthetic data with exponential noise | PI quality | Helped |
| Use the distribution-free QD loss rather than Gaussian-assumption MVE (ablation) | Neural-network prediction intervals · Synthetic data with normal noise | PI quality | Hurt |
| Add an ensemble to a single QD network (ablation) | QD prediction intervals · Synthetic regression data with sparsely represented input regions | model uncertainty estimation | Helped |
| Use bootstrap resampling rather than parameter resampling (ablation) | QD-Ens · Ten benchmark regression datasets | PI quality | Hurt |
Closest papers by meaning
- Building Calibrated Deep Models via Uncertainty Matching with Auxiliary Interval Predictors (AAAI 2020 · similarity 0.90)
- Simple and Scalable Predictive Uncertainty Estimation using Deep Ensembles (NeurIPS 2017 · similarity 0.89)
- Locally Valid and Discriminative Prediction Intervals for Deep Learning Models (NeurIPS 2021 · similarity 0.89)
- Uncertainty Quantification for Deep Regression using Contextualised Normalizing Flows (NeurIPS 2025 · similarity 0.88)
- Approximating Full Conformal Prediction for Neural Network Regression with Gauss-Newton Influence (ICLR 2025 · similarity 0.87)
- SURE: SUrvey REcipes for building reliable and robust deep networks (CVPR 2024 · similarity 0.86)
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.