Learning Clean Generative Processes from Noisy Measurements
arXiv cs.AI · 31 Oct 2026, 03:01 AEST
Researchers explore learning generative models of clean images from noisy observations. They demonstrate that consistency in predictions can enhance the generation of clean images from various noise levels.
Key facts
- We study how to learn a generative model of clean images using only noisy observations.
- To learn the predictions below that level, needed to generate clean images, we impose consistency: a prediction should equal its expectation after continuing the sampling process.
- Experiments on CIFAR-10, photon-counting microscopy, and synthetic aperture radar show that the required posterior predictions can be learned from noisy views alone and that the consistency relation repairs the continuation beyond the noise level of the data, which severely deteriorates without consistency.
Read aloud
Researchers explore learning generative models of clean images from noisy observations. They demonstrate that consistency in predictions can enhance the generation of clean images from various noise levels. We study how to learn a generative model of clean images using only noisy observations. To learn the predictions below that level, needed to generate clean images, we impose consistency: a prediction should equal its expectation after continuing the sampling process. Experiments on CIFAR-ten, photon-counting microscopy, and synthetic aperture radar show that the required posterior predictions can be learned from noisy views alone and that the consistency relation repairs the continuation beyond the noise level of the data, which severely deteriorates without consistency.
Sources in this cluster
- Learning Clean Generative Processes from Noisy Measurements
arXiv cs.AI · 31 Oct 2026, 03:01 AEST