analiza aiqbench
This research introduces DDA-UQ, a framework that enhances the reliability of Vision-Language Models by dynamically adapting uncertainty estimates to shifting data distributions rather than relying on static mappings. By leveraging Gaussian Mixture Models during training, the method effectively addresses domain shift issues critical for deploying AI in safety-sensitive applications. The approach demonstrates superior performance over existing post-hoc techniques, offering a more robust safeguard against model failure in real-world scenarios.
Dynamic Distribution-Aware Uncertainty Tracking in Vision-Language Representation Learning
arXiv:2608.09011v2 Announce Type: replace Abstract: Uncertainty Quantification (UQ) aims to measure the reliability of model predictions, serving as a critical safeguard for deploying Vision-Language Models (VLMs) in safety-critical scenarios. Post-hoc approaches are widely adopted due to their lightweight nature, mapping the outputs of VLMs to uncertainty measures through learnable modules or inductive summarization. However, Post-hoc approaches remain inherently confined to fitting the failure patterns of the source domain, ignoring the dynamic nature of test distributions. To address this challenge, we propose a Dynamic Distribution-Aware Uncertainty Quantification framework (DDA-UQ) that shifts the paradigm from static mapping to a dynamic distribution-aware process. During training, we leverage a Gaussian Mixture Model to model the VVLMs'embedding space and extract distributional evidence, thereby dynamically deriving uncertainty estimates. During inference, the design dynamically responds to changes in the data distribution. Extensive experiments demonstrate that our approach significantly outperforms state-of-the-art methods.
Abstract:Uncertainty Quantification (UQ) aims to measure the reliability of model predictions, serving as a critical safeguard for deploying Vision-Language Models (VLMs) in safety-critical scenarios. Post-hoc approaches are widely adopted due to their lightweight nature, mapping the outputs of VLMs to uncertainty measures through learnable modules or inductive summarization. However, Post-hoc approaches remain inherently confined to fitting the failure patterns of the source domain, ignoring the dynamic nature of test distributions. To address this challenge, we propose a Dynamic Distribution-Aware Uncertainty Quantification framework (DDA-UQ) that shifts the paradigm from static mapping to a dynamic distribution-aware process. During training, we leverage a Gaussian Mixture Model to model the VVLMs'embedding space and extract distributional evidence, thereby dynamically deriving uncertainty estimates. During inference, the design dynamically responds to changes in the data distribution. Extensive experiments demonstrate that our approach significantly outperforms state-of-the-art methods.
Machine Learning (cs.LG)
arXiv:2608.09011 [cs.LG]
(or arXiv:2608.09011v2 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2608.09011 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 231 cuvinte
Subiecte conexe
BRICS Draws a New Global Line: Full Text of New Delhi Declaration
slguardian.org · 12 sept. 2026
D3O: Dynamic Distribution Distillation for Ordinal Regression
arXiv · 28 iul. 2026
Always ask your agent to be honest
LessWrong · 16 sept. 2026
2026 BAIR Graduate Showcase
BAIR Blog · 1 iul. 2026