BayesDiff：通过贝叶斯推理估计扩散模型中的像素级不确定性
BayesDiff: Estimating Pixel-wise Uncertainty in Diffusion via Bayesian Inference
AI总结：
本文提出 BayesDiff，利用贝叶斯推理、不确定性迭代原则和最后一层 Laplace 近似估计扩散生成图像的像素级不确定性，可用于筛选低质量样本、增强成功生成并修复伪影。
摘要：
扩散模型具有出色的图像生成能力，但低质量生成结果仍然存在，并且由于缺乏合适的逐样本指标，识别这些结果仍具挑战性。为解决该问题，我们提出 BayesDiff，这是一种基于贝叶斯推理、用于扩散模型生成结果的像素级不确定性估计器。具体而言，我们推导了一种新颖的不确定性迭代原则，以刻画扩散过程中的不确定性动态，并利用最后一层 Laplace 近似实现高效的贝叶斯推理。估计得到的像素级不确定性不仅可以聚合为逐样本指标，以过滤低保真图像，还能在文本到图像任务中帮助增强成功生成，并修正失败生成中的伪影。大量实验证明了 BayesDiff 的有效性及其在实际应用中的前景。
英文摘要：
Diffusion models have impressive image generation capability, but low-quality generations still exist, and their identification remains challenging due to the lack of a proper sample-wise metric. To address this, we propose BayesDiff, a pixel-wise uncertainty estimator for generations from diffusion models based on Bayesian inference. In particular, we derive a novel uncertainty iteration principle to characterize the uncertainty dynamics in diffusion, and leverage the last-layer Laplace approximation for efficient Bayesian inference. The estimated pixel-wise uncertainty can not only be aggregated into a sample-wise metric to filter out low-fidelity images but also aids in augmenting successful generations and rectifying artifacts in failed generations in text-to-image tasks. Extensive experiments demonstrate the efficacy of BayesDiff and its promise for practical applications.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
