analiza aiqbench
Posterior Flow Matching introduces a generative framework that simultaneously learns transport predictions and endpoint uncertainty by modeling conditional velocity means and variances. This approach enables adaptive noise scaling in both ODE and SDE formulations, offering improved generation quality across diverse domains like images and molecular structures while providing diagnostic readouts throughout the process. The method addresses a key limitation in standard flow matching where intermediate uncertainty is not directly reported, enhancing reliability for complex data synthesis tasks.
Posterior Flow Matching
arXiv:2610.13994v1 Announce Type: cross Abstract: Flow matching learns how to transport noise toward data, but its mean prediction does not directly report the endpoint uncertainty remaining at an intermediate state. This paper introduces Posterior Flow Matching (PFM), a general generative modeling framework that learns conditional velocity means and componentwise variances with a Gaussian posterior model. The same model evaluation supplies both a transport prediction and endpoint uncertainty, making posterior diagnostics available throughout generation. These learned moments support a mean-driven ordinary differential equation (ODE) and a posterior-driven stochastic differential equation (SDE), whose local noise scales adapt to the predicted uncertainty. Across image, point-cloud, molecular, and action generation, PFM improves generation quality and provides endpoint-posterior readouts throughout generation.
Abstract:Flow matching learns how to transport noise toward data, but its mean prediction does not directly report the endpoint uncertainty remaining at an intermediate state. This paper introduces Posterior Flow Matching (PFM), a general generative modeling framework that learns conditional velocity means and componentwise variances with a Gaussian posterior model. The same model evaluation supplies both a transport prediction and endpoint uncertainty, making posterior diagnostics available throughout generation. These learned moments support a mean-driven ordinary differential equation (ODE) and a posterior-driven stochastic differential equation (SDE), whose local noise scales adapt to the predicted uncertainty. Across image, point-cloud, molecular, and action generation, PFM improves generation quality and provides endpoint-posterior readouts throughout generation.
Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
arXiv:2610.13994 [cs.LG]
(or arXiv:2610.13994v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2610.13994 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
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 · 214 cuvinte
Subiecte conexe
Posterior-Averaged Bellman Flow Matching for Distributional Critics
arXiv · 15 oct. 2026
Flow Matching Calibration for Simulation-Based Inference under Model Misspecification
arXiv · 5 iun. 2026
Getting Your Guidance Weights Right in diffusion and flow-matching posterior sampling
arXiv · 5 oct. 2026
Gaussian Flow Dynamics: Simulation-Free Neural SDE Learning Beyond One-Time Marginals
arXiv · 6 oct. 2026