Saved in:
Bibliographic Details
Main Authors: Yair, Omer, Nehme, Elias, Michaeli, Tomer
Format: Preprint
Published: 2023
Subjects:
Online Access:https://arxiv.org/abs/2312.07804
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910443759468544
author Yair, Omer
Nehme, Elias
Michaeli, Tomer
author_facet Yair, Omer
Nehme, Elias
Michaeli, Tomer
contents In ill-posed inverse problems, it is commonly desirable to obtain insight into the full spectrum of plausible solutions, rather than extracting only a single reconstruction. Information about the plausible solutions and their likelihoods is encoded in the posterior distribution. However, for high-dimensional data, this distribution is challenging to visualize. In this work, we introduce a new approach for estimating and visualizing posteriors by employing energy-based models (EBMs) over low-dimensional subspaces. Specifically, we train a conditional EBM that receives an input measurement and a set of directions that span some low-dimensional subspace of solutions, and outputs the probability density function of the posterior within that space. We demonstrate the effectiveness of our method across a diverse range of datasets and image restoration problems, showcasing its strength in uncertainty quantification and visualization. As we show, our method outperforms a baseline that projects samples from a diffusion-based posterior sampler, while being orders of magnitude faster. Furthermore, it is more accurate than a baseline that assumes a Gaussian posterior.
format Preprint
id arxiv_https___arxiv_org_abs_2312_07804
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Uncertainty Visualization via Low-Dimensional Posterior Projections
Yair, Omer
Nehme, Elias
Michaeli, Tomer
Computer Vision and Pattern Recognition
In ill-posed inverse problems, it is commonly desirable to obtain insight into the full spectrum of plausible solutions, rather than extracting only a single reconstruction. Information about the plausible solutions and their likelihoods is encoded in the posterior distribution. However, for high-dimensional data, this distribution is challenging to visualize. In this work, we introduce a new approach for estimating and visualizing posteriors by employing energy-based models (EBMs) over low-dimensional subspaces. Specifically, we train a conditional EBM that receives an input measurement and a set of directions that span some low-dimensional subspace of solutions, and outputs the probability density function of the posterior within that space. We demonstrate the effectiveness of our method across a diverse range of datasets and image restoration problems, showcasing its strength in uncertainty quantification and visualization. As we show, our method outperforms a baseline that projects samples from a diffusion-based posterior sampler, while being orders of magnitude faster. Furthermore, it is more accurate than a baseline that assumes a Gaussian posterior.
title Uncertainty Visualization via Low-Dimensional Posterior Projections
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2312.07804