Entity · paper

Subjective Risk Decomposition: A New View for Uncertainty Quantification

paperactiveprovisionalsubjective-risk-decomposition-a-new-view-for-uncertainty-quantification-31894ef7·1 events·first seen Jul 17, 2026

Aliases: Subjective Risk Decomposition: A New View for Uncertainty Quantification

More like this (12)

Uncertainty Quantification Invariant Risk Minimization BeyondUncertainty Beyond Aggregate Risk: Role-Stratified Conformal Risk Control for LLM Tool Calls Uncertainty Quantification for Computer-Use Agents: A Benchmark across Vision-Language Models and GUI Grounding Datasets Uncertainty Calibration Uncertainty-aware Multi-Granularity RAG Uncertainty-Aware Generation and Decision-Making Under Ambiguity Just how sure are you? Improving Verbalized Uncertainty Calibration in Medical VQA Conformal Risk Control Task Decomposition Hierarchical Analysis Framework Risk Direction Index

Recent events (1)

5arXiv · cs.AI·Jul 17, 2026·source ↗

Subjective Risk Decomposition: A unified theoretical framework for uncertainty quantification

A new arXiv preprint proposes deriving epistemic and aleatoric uncertainty measures as consequences of decomposing a subjective risk based on a strictly proper loss, rather than treating them as axiomatic primitives. The framework unifies numerous previously proposed UQ measures under a common theoretical foundation, with reverse cross-entropy recovering classic information-theoretic uncertainty terms as a special case. The authors extend the view to learning theory, introducing subjective risk analogues of excess risk, approximation error, and estimation error, positioning this as a first step toward a full learning-theoretic framework for UQ.

Evaluation and Benchmarking AI Safety Research Subjective Risk Decomposition: A New View for Uncertainty Quantification