Search papers, labs, and topics across Lattice.
This paper introduces CREDENCE, a concept bottleneck model framework that decomposes concept uncertainty into epistemic and aleatoric components by representing concepts as credal predictions (probability intervals). Epistemic uncertainty is derived from disagreement across diverse concept heads, while aleatoric uncertainty is estimated via a dedicated ambiguity output trained to match annotator disagreement. Experiments across several tasks demonstrate that epistemic uncertainty correlates with prediction errors, while aleatoric uncertainty tracks annotator disagreement, enabling prescriptive decision-making.
Concept bottleneck models can now distinguish between reducible model uncertainty and irreducible input ambiguity, enabling targeted interventions like data collection and human review.
Concept Bottleneck Models (CBMs) predict through human-interpretable concepts, but they typically output point concept probabilities that conflate epistemic uncertainty (reducible model underspecification) with aleatoric uncertainty (irreducible input ambiguity). This makes concept-level uncertainty hard to interpret and, more importantly, hard to act upon. We introduce CREDENCE (Credal Ensemble Concept Estimation), a CBM framework that decomposes concept uncertainty by construction. CREDENCE represents each concept as a credal prediction (a probability interval), derives epistemic uncertainty from disagreement across diverse concept heads, and estimates aleatoric uncertainty via a dedicated ambiguity output trained to match annotator disagreement when available. The resulting signals support prescriptive decisions: automate low-uncertainty cases, prioritize data collection for high-epistemic cases, route high-aleatoric cases to human review, and abstain when both are high. Across several tasks, we show that epistemic uncertainty is positively associated with prediction errors, whereas aleatoric uncertainty closely tracks annotator disagreement, providing guidance beyond error correlation. Our implementation is available at the following link: https://github.com/Tankiit/Credal_Sets/tree/ensemble-credal-cbm
