Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Instance-Based Uncertainty Estimation for Gradient-Boosted Regression Trees
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 IBUG instead of NGBoost | Probabilistic regression on 22 benchmark datasets · 22 regression datasets | CRPS head-to-head wins: 17-3-2 | Helped |
| Use IBUG instead of PGBM | Probabilistic regression on 22 benchmark datasets · 22 regression datasets | CRPS head-to-head wins: 11-9-2 | Helped |
| Average IBUG and CBU predictions | IBUG · 22 regression datasets | CRPS head-to-head wins against IBUG: 15-6-1 | Mixed |
| Use LightGBM instead of CatBoost as the IBUG base model (ablation) | IBUG · Bike | CRPS: 0.974 -> 0.819 | Helped |
| Use Weibull fitted to neighbors instead of a normal distribution (ablation) | IBUG · MEPS | NLL: 3.699 -> -6.887 | Helped |
| Use KDE instead of a normal distribution (ablation) | IBUG · Wine | NLL: 1.025 -> 0.785 | Helped |
| Add validation-tuned affine variance calibration (ablation) | IBUG, NGBoost, PGBM, and CBU · 22 regression datasets | CRPS: IBUG 13 wins, 9 ties, 0 losses; CBU 17 wins, 5 ties, 0 losses | Helped |
| Sample a subset of the first trees when computing affinities (ablation) | IBUG · Six datasets | prediction time: 1-2 orders of magnitude | Helped |
Closest papers by meaning
- Smooth And Consistent Probabilistic Regression Trees (NeurIPS 2020 · similarity 0.88)
- Statistical Inference for Gradient Boosting Regression (NeurIPS 2025 · similarity 0.86)
- Regression Trees Know Calculus (NeurIPS 2025 · similarity 0.86)
- Treeffuser: probabilistic prediction via conditional diffusions with gradient-boosted trees (NeurIPS 2024 · similarity 0.85)
- GRANDE: Gradient-Based Decision Tree Ensembles for Tabular Data (ICLR 2024 · similarity 0.84)
- Gradient Regularized Newton Boosting Trees with Global Convergence (NeurIPS 2026 · 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.