Research map / Learning theory and algorithms / Prediction and decision losses
Explaining Predictive Uncertainty with Information Theoretic Shapley Values
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 |
|---|---|---|---|
| rank features by absolute entropy Shapley values under 50% missing data | Entropy Shapley values · Active feature-value acquisition (modified Friedman) | AUC of feature importance ranking: AUC 0.682 | Helped |
| increase missingness from 0 to 50 percent | Entropy Shapley values · Active feature-value acquisition (modified Friedman) | AUC of feature importance ranking | Hurt |
| use split-conformal intervals on Shapley values | Entropy Shapley values · Modified Friedman | empirical coverage at alpha=0.1: 0.895 to 0.904 vs target 0.900 | Helped |
| explain entropy (variance model) instead of conditional mean predictions | Shapley values · Modified Friedman | feature attribution concentration near zero | Helped |
| vary conditional sampler (KernelSHAP, ML, copula, empirical, ctree) | Shapley value estimation for entropy game · Gaussian simulation with Toeplitz covariance | mean absolute error | Mixed |
Closest papers by meaning
- Incorporating Information into Shapley Values: Reweighting via a Maximum Entropy Approach (ICML 2024 · similarity 0.88)
- Regression-adjusted Monte Carlo Estimators for Shapley Values and Probabilistic Values (NeurIPS 2025 · similarity 0.87)
- Explaining Probabilistic Models with Distributional Values (ICML 2024 · similarity 0.86)
- Rethinking Aleatoric and Epistemic Uncertainty (ICML 2025 · similarity 0.86)
- shapiq: Shapley Interactions for Machine Learning (NeurIPS 2024 · similarity 0.86)
- An Information Theoretic Perspective on Conformal Prediction (NeurIPS 2024 · similarity 0.85)
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.