Diffusion Prior-Based Amortized Variational Inference for Noisy Inverse Problems

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Lee, Sojin, Park, Dogyun, Kong, Inho, Kim, Hyunwoo J.
Format: Preprint
Veröffentlicht: 2024
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866913441896202240
author Lee, Sojin
Park, Dogyun
Kong, Inho
Kim, Hyunwoo J.
author_facet Lee, Sojin
Park, Dogyun
Kong, Inho
Kim, Hyunwoo J.
contents Recent studies on inverse problems have proposed posterior samplers that leverage the pre-trained diffusion models as powerful priors. These attempts have paved the way for using diffusion models in a wide range of inverse problems. However, the existing methods entail computationally demanding iterative sampling procedures and optimize a separate solution for each measurement, which leads to limited scalability and lack of generalization capability across unseen samples. To address these limitations, we propose a novel approach, Diffusion prior-based Amortized Variational Inference (DAVI) that solves inverse problems with a diffusion prior from an amortized variational inference perspective. Specifically, instead of separate measurement-wise optimization, our amortized inference learns a function that directly maps measurements to the implicit posterior distributions of corresponding clean data, enabling a single-step posterior sampling even for unseen measurements. Extensive experiments on image restoration tasks, e.g., Gaussian deblur, 4$\times$ super-resolution, and box inpainting with two benchmark datasets, demonstrate our approach's superior performance over strong baselines. Code is available at https://github.com/mlvlab/DAVI.
format Preprint
id arxiv_https___arxiv_org_abs_2407_16125
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Diffusion Prior-Based Amortized Variational Inference for Noisy Inverse Problems
Lee, Sojin
Park, Dogyun
Kong, Inho
Kim, Hyunwoo J.
Computer Vision and Pattern Recognition
Recent studies on inverse problems have proposed posterior samplers that leverage the pre-trained diffusion models as powerful priors. These attempts have paved the way for using diffusion models in a wide range of inverse problems. However, the existing methods entail computationally demanding iterative sampling procedures and optimize a separate solution for each measurement, which leads to limited scalability and lack of generalization capability across unseen samples. To address these limitations, we propose a novel approach, Diffusion prior-based Amortized Variational Inference (DAVI) that solves inverse problems with a diffusion prior from an amortized variational inference perspective. Specifically, instead of separate measurement-wise optimization, our amortized inference learns a function that directly maps measurements to the implicit posterior distributions of corresponding clean data, enabling a single-step posterior sampling even for unseen measurements. Extensive experiments on image restoration tasks, e.g., Gaussian deblur, 4$\times$ super-resolution, and box inpainting with two benchmark datasets, demonstrate our approach's superior performance over strong baselines. Code is available at https://github.com/mlvlab/DAVI.
title Diffusion Prior-Based Amortized Variational Inference for Noisy Inverse Problems
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2407.16125