analiza aiqbench
This paper introduces UA-Flow, a method that integrates uncertainty quantification directly into the flow matching framework to assess sample reliability and steer generation quality. By propagating velocity uncertainty through flow dynamics, the approach provides spatially localized signals that correlate more consistently with data-manifold support than standard element-wise baselines. The results demonstrate improved fidelity metrics on ImageNet and CIFAR-10 while offering new guidance strategies for balancing precision and recall in generative modeling.
Flow Matching with Uncertainty Quantification and Guidance
arXiv:2602.10326v3 Announce Type: replace-cross Abstract: Despite the remarkable success of sampling-based generative models such as flow matching, they can still produce samples of inconsistent or degraded quality. To assess sample reliability and improve sample fidelity, we propose uncertainty-aware flow matching (UA-Flow), a lightweight extension of flow matching that predicts the velocity field together with heteroscedastic uncertainty. UA-Flow estimates spatially localized sample uncertainty by propagating velocity uncertainty through the flow dynamics. The resulting estimates serve as reliability signals that reflect data-manifold support for each generated sample or region. We further use them to steer generation via uncertainty-aware guidance (UG) and uncertainty-aware classifier-free guidance (UCFG). Experiments on ImageNet and CIFAR-10 show that UA-Flow's uncertainty tracks sample fidelity more consistently than element-wise baselines and that these signals also carry information about data-manifold support. On ImageNet, UG increases precision at the cost of recall and improves FID, while UCFG maintains lower FID than CFG at large guidance scales.
Abstract:Despite the remarkable success of sampling-based generative models such as flow matching, they can still produce samples of inconsistent or degraded quality. To assess sample reliability and improve sample fidelity, we propose uncertainty-aware flow matching (UA-Flow), a lightweight extension of flow matching that predicts the velocity field together with heteroscedastic uncertainty. UA-Flow estimates spatially localized sample uncertainty by propagating velocity uncertainty through the flow dynamics. The resulting estimates serve as reliability signals that reflect data-manifold support for each generated sample or region. We further use them to steer generation via uncertainty-aware guidance (UG) and uncertainty-aware classifier-free guidance (UCFG). Experiments on ImageNet and CIFAR-10 show that UA-Flow's uncertainty tracks sample fidelity more consistently than element-wise baselines and that these signals also carry information about data-manifold support. On ImageNet, UG increases precision at the cost of recall and improves FID, while UCFG maintains lower FID than CFG at large guidance scales.
Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
arXiv:2602.10326 [cs.CV]
(or arXiv:2602.10326v3 [cs.CV] for this version)
https://doi.org/10.48550/arXiv.2602.10326 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 · 247 cuvinte
Subiecte conexe
PosteriorBench: From Point Estimates to Posterior Matching in Evaluating Generative Inverse Solvers
arXiv · 18 sept. 2026
Uncertainty Quantification for Flow-Based Vision-Language-Action Models
arXiv · 17 iun. 2026
GeoFunFlow: Geometric function flow matching for joint probabilistic inference of physical fields and complex geometries
arXiv · 6 oct. 2026
Deterministic Prediction as Global Velocity: Flow Matching for Probabilistic Time-Series Forecast
arXiv · 14 oct. 2026