How Good Are Deep Generative Models for Solving Inverse Problems?
On the map: Diffusion models → Inverse problems and image restoration
Abstract (arXiv)
Deep generative models, such as diffusion models, GANs, and IMLE, have shown impressive capability in tackling inverse problems. However, the validity of model-generated solutions w.r.t. the forward problem and the reliability of associated uncertainty estimates remain understudied. This study evaluates recent diffusion-based, GAN-based, and IMLE-based methods on three inverse problems, i.e., 16× super-resolution, colourization, and image decompression. We assess the validity of these models' outputs as solutions to the inverse problems and conduct a thorough analysis of the reliability of the models' estimates of uncertainty over the solution. Overall, we find that the IMLE-based CHIMLE method outperforms other methods in terms of producing valid solutions and reliable uncertainty estimates.
Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Placed in Inverse problems and image restoration, one of 14 directions in Diffusion models. Builds on 24 of its 27 references that are in the map. First posted on arXiv 20 Dec 2023.
Builds on
24 of 27 references.
Closest ancestors · 1
Same direction, newest first
- 2022Denoising Diffusion Restoration Models27 Jan
Other references · 23
- 2020Denoising Diffusion Probabilistic Modelscited by 21,076
- 2018The Unreasonable Effectiveness of Deep Features as a Perceptual Metriccited by 11,177
- 2014ImageNet Large Scale Visual Recognition Challengecited by 8,498
- 2015Deep Unsupervised Learning using Nonequilibrium Thermodynamicscited by 7,058
- 2016Image-to-Image Translation with Conditional Adversarial Networkscited by 4,527
- 2021A Gentle Introduction to Conformal Prediction and Distribution-Free Uncertainty Quantificationcited by 792
- 2017Toward Multimodal Image-to-Image Translationcited by 296
- 2021Distribution-Free, Risk-Controlling Prediction Setscited by 288
- 2019Mode Seeking Generative Adversarial Networks for Diverse Image Synthesiscited by 138
- 2022Image-to-Image Regression with Distribution-Free Uncertainty Quantification and Applications in Imagingcited by 107
- 2018Implicit Maximum Likelihood Estimationcited by 52
- 2023How to Trust Your Diffusion Model: A Convex Optimization Approach to Conformal Risk Controlcited by 50
11 more
- 2022QS-Attn: Query-Selected Attention for Contrastive Learning in I2I Translationcited by 39
- 2021DivCo: Diverse Conditional Image Synthesis via Contrastive Generative Adversarial Networkcited by 37
- 2022Predictive Inference with Feature Conformal Predictioncited by 34
- 2022Conffusion: Confidence Intervals for Diffusion Modelscited by 33
- 2020Image Colorization: A Survey and Datasetcited by 23
- 2022Modulated Contrast for Versatile Image Synthesiscited by 23
- 2022Non-Uniform Diffusion Modelscited by 14
- 2020Multimodal Image Synthesis with Conditional Implicit Maximum Likelihood Estimationcited by 13
- 2015Fast k-Nearest Neighbour Search via Dynamic Continuous Indexingcited by 6
- 2017Fast k-Nearest Neighbour Search via Prioritized DCIcited by 5
- 2022CHIMLE: Conditional Hierarchical IMLE for Multimodal Conditional Image Synthesiscited by 4
Built on by
Citing papers per year: 2025: 1.
1 paper.
Similar, not cited
Close in wording, no citation either way
- 2024A Survey on Diffusion Models for Inverse Problems
- 2021Differentiable Gaussianization Layers for Inverse Problems Regularized by Deep Generative Models
- 2025Image Inversion: A Survey from GANs to Diffusion and Beyond
- 2023A Variational Perspective on Solving Inverse Problems with Diffusion Models
- 2023Solving Inverse Problems with Latent Diffusion Models via Hard Data Consistency
- 2024Comparative Analysis of Generative Models: Enhancing Image Synthesis with VAEs, GANs, and Stable Diffusion
- 2023Training-free Linear Image Inverses via Flows
- 2024Convergence Properties of Score-Based Models for Linear Inverse Problems Using Graduated Optimisation