#uncertainty-quantification3 articles

3 articles

AIBullisharXiv – CS AI · 4h ago2

🧠

Uncertainty Quantification for Multimodal Large Language Models with Incoherence-adjusted Semantic Volume

Researchers introduce UMPIRE, a new training-free framework for quantifying uncertainty in Multimodal Large Language Models (MLLMs) across various input and output modalities. The system measures incoherence-adjusted semantic volume of model responses to better detect errors and improve reliability without requiring external tools or additional computational overhead.

AIBullisharXiv – CS AI · 4h ago2

🧠

Evidential Neural Radiance Fields

Researchers introduce Evidential Neural Radiance Fields, a new probabilistic approach that enables uncertainty quantification in 3D scene modeling while maintaining rendering quality. The method addresses critical limitations in existing NeRF technology by capturing both aleatoric and epistemic uncertainty from a single forward pass, making neural radiance fields more suitable for safety-critical applications.

AINeutralarXiv – CS AI · 4h ago2

🧠

RewardUQ: A Unified Framework for Uncertainty-Aware Reward Models

Researchers introduce RewardUQ, a unified framework for evaluating uncertainty quantification in reward models used to align large language models with human preferences. The study finds that model size and initialization have the most significant impact on performance, while providing an open-source Python package to advance the field.