aiqbench analysis
This research advances generative modeling for physical fields by training Bayesian Flow Networks to learn the distribution of authentic, structured occlusions rather than relying on heuristic masking strategies. By aligning mask priors with real-world observation topologies, the method effectively prevents generative collapse and improves reconstruction accuracy on sparse satellite data. The approach represents a significant step toward robust AI systems capable of handling incomplete physical observations in complex environments like oceanography.
Observation-Aligned Mask Priors for Learning Physical Fields from Authentic Occlusions
arXiv:2605.16818v2 Announce Type: replace-cross Abstract: Learning physical fields directly from incomplete observations is challenging because authentic occlusions are structured, sample-dependent, and often missing not at random, whereas existing methods typically rely on heuristic masking rules or predefined mask distributions. We propose Observation-Aligned Mask Priors, a framework that learns the distribution of authentic observation masks and uses it to construct context-query partitions for training from incomplete data. Specifically, we pretrain a Bayesian Flow Network (BFN) on binary observation masks to capture real occlusion topologies, then guide BFN sampling with a globally normalized cross-entropy objective to generate sample-specific masks aligned with each sparse observation. The intersection between the guided mask and the observed mask defines the context, and the remaining observed entries become query targets for a diffusion-based reconstruction model. We show that this intersection-based partitioning gives every valid observed dimension a strictly positive probability of being queried, preventing zero-query dead zones and local generative collapse. Experiments on three real-world oceanographic datasets with authentic satellite occlusions, across resolutions up to 256$\times$256, show consistent improvements over strong diffusion baselines in MSE and PSNR. These results demonstrate that learning mask priors from authentic occlusions is an effective alternative to heuristic masking for learning from incomplete physical observations without access to fully observed fields.
Abstract:Learning physical fields directly from incomplete observations is challenging because authentic occlusions are structured, sample-dependent, and often missing not at random, whereas existing methods typically rely on heuristic masking rules or predefined mask distributions. We propose Observation-Aligned Mask Priors, a framework that learns the distribution of authentic observation masks and uses it to construct context-query partitions for training from incomplete data. Specifically, we pretrain a Bayesian Flow Network (BFN) on binary observation masks to capture real occlusion topologies, then guide BFN sampling with a globally normalized cross-entropy objective to generate sample-specific masks aligned with each sparse observation. The intersection between the guided mask and the observed mask defines the context, and the remaining observed entries become query targets for a diffusion-based reconstruction model. We show that this intersection-based partitioning gives every valid observed dimension a strictly positive probability of being queried, preventing zero-query dead zones and local generative collapse. Experiments on three real-world oceanographic datasets with authentic satellite occlusions, across resolutions up to 256$\times$256, show consistent improvements over strong diffusion baselines in MSE and PSNR. These results demonstrate that learning mask priors from authentic occlusions is an effective alternative to heuristic masking for learning from incomplete physical observations without access to fully observed fields.
Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
arXiv:2605.16818 [cs.CV]
(or arXiv:2605.16818v2 [cs.CV] for this version)
https://doi.org/10.48550/arXiv.2605.16818 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.
Full text under the publisher's open licence · 302 words
Related coverage
The Machines that Make the Machines
NVIDIA · Oct 7, 2026
Generalization and infinite width
LessWrong · Aug 1, 2026
The Alignment Problem: Computer Science Meets Analytic Philosophy
LessWrong · Oct 18, 2026
Humanity’s Last Invention — Richard Socher of Recursive
Latent Space · Sep 14, 2026