从噪声观测中学习干净的生成过程
Learning Clean Generative Processes from Noisy Measurements
AI 导读
一项 arXiv 研究提出仅用带噪声的观测学习干净图像的扩散生成模型。方法先用 Noise2Noise 回归从独立噪声视图学习后验预测，但仅适用于不低于训练噪声水平的范围；为此引入一致性约束——预测应等于继续采样后的期望——并推导了该约束唯一决定干净图像分布延续的条件，在 Poisson、binomial、Gamma 与 negative-binomial 测量下验证成立。
正文 · 原文
Abstract:We study how to learn a generative model of clean images using only noisy observations. Diffusion processes and their generalizations to non-Gaussian noise require posterior predictions at every noise level. We first show that it is possible to learn these predictions from independent noisy views using Noise2Noise regression, but only at or above the noise level of the training data. 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. We derive conditions under which consistency uniquely determines the continuation to the clean image distribution and verify them for Poisson, binomial, Gamma, and negative-binomial measurements. 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.
| Subjects: | Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG) |
| Cite as: | arXiv:2610.42036 [cs.CV] |
| (or arXiv:2610.42036v1 [cs.CV] for this version) | |
| https://doi.org/10.48550/arXiv.2610.42036 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Fabian Kruse [view email]
[v1]
Fri, 30 Oct 2026 17:01:25 UTC (994 KB)
来源：arXiv cs.AI · arxiv.org