U-Nets as Belief Propagation: Efficient Classification, Denoising, and Diffusion in Generative Hierarchical Models

arXiv:2404.18444v1 Announce Type: new
Abstract: U-Nets are among the most widely used architectures in computer vision, renowned for their exceptional performance in applications such as image segmentation, denoising, and diffusion modeling. However, a theoretical explanation of the U-Net architecture design has not yet been fully established.
This paper introduces a novel interpretation of the U-Net architecture by studying certain generative hierarchical models, which are tree-structured graphical models extensively utilized in both language and image domains. With their encoder-decoder structure, long …

abstract applications architecture architectures arxiv belief classification computer computer vision cs.ai cs.lg denoising design diffusion diffusion modeling generative hierarchical however image math.st modeling performance propagation segmentation stat.ml stat.th type vision