CREDO: Epistemic-Aware Conformalized Credal Envelopes for Regression

2026-03-06

AI Summary

Standard conformal prediction often exhibits overconfidence in model extrapolation regions due to its neglect of epistemic uncertainty, while traditional credible sets, though capturing such uncertainty, lack calibration guarantees. To address this, this work proposes CREDO: a method that first constructs interpretable credible envelopes whose width increases as local evidence diminishes, and then applies split conformal calibration on these envelopes to achieve marginal coverage guarantees without additional assumptions. CREDO uniquely decouples and integrates credible sets with conformal prediction, decomposing prediction interval width into three interpretable components—aleatoric noise, epistemic inflation, and distribution-free calibration slack—thereby offering both theoretical guarantees and interpretability. Experiments across multiple regression benchmarks demonstrate that CREDO maintains target coverage while significantly improving adaptivity in data-sparse regions, all with competitive computational efficiency.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Reasoning under Uncertainty: Other Foundations of Reasoning under Uncertainty
• Computer Vision: Interpretability, Explainability, and Transparency

Application Category

• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Conformal prediction delivers prediction intervals with distribution-free coverage, but its intervals can look overconfident in regions where the model is extrapolating, because standard conformal scores do not explicitly represent epistemic uncertainty. Credal methods, by contrast, make epistemic effects visible by working with sets of plausible predictive distributions, but they are typically model-based and lack calibration guarantees. We introduce CREDO, a simple"credal-then-conformalize"recipe that combines both strengths. CREDO first builds an interpretable credal envelope that widens when local evidence is weak, then applies split conformal calibration on top of this envelope to guarantee marginal coverage without further assumptions. This separation of roles yields prediction intervals that are interpretable: their width can be decomposed into aleatoric noise, epistemic inflation, and a distribution-free calibration slack. We provide a fast implementation based on trimming extreme posterior predictive endpoints, prove validity, and show on benchmark regressions that CREDO maintains target coverage while improving sparsity adaptivity at competitive efficiency.

Problem

• conformal prediction
• epistemic uncertainty
• credal sets
• regression
• prediction intervals

Innovation

• conformal prediction
• credal sets
• epistemic uncertainty
• prediction intervals
• calibration