Corpus ID: 211258601

Tuning-free Plug-and-Play Proximal Algorithm for Inverse Imaging Problems

@inproceedings{Wei2020TuningfreePP,
  title={Tuning-free Plug-and-Play Proximal Algorithm for Inverse Imaging Problems},
  author={Kaixuan Wei and Angelica I. Avil{\'e}s-Rivero and Jingwei Liang and Ying Fu and C. Sch{\"o}nlieb and H. Huang},
  booktitle={ICML},
  year={2020}
}
Plug-and-play (PnP) is a non-convex framework that combines ADMM or other proximal algorithms with advanced denoiser priors. Recently, PnP has achieved great empirical success, especially with the integration of deep learning-based denoisers. However, a key problem of PnP based approaches is that they require manual parameter tweaking. It is necessary to obtain high-quality results across the high discrepancy in terms of imaging conditions and varying scene content. In this work, we present a… Expand
Scalable Plug-and-Play ADMM With Convergence Guarantees
TLDR
This work theoretically analyze the convergence of the PnP-ADMM algorithm under a set of explicit assumptions, extending recent theoretical results in the area, and shows the effectiveness of the algorithm with nonsmooth data-fidelity terms and deep neural net priors, its fast convergence compared to existing PNP algorithms, and its scalability in terms of speed and memory. Expand
Distributed optimization with tunable learned priors for robust ptycho-tomography
TLDR
This work extends the ADMM framework with plug-and-play (PnP) denoisers by replacing the regularization subproblem with a general denoising operator based on machine learning, and proposes a tuning parameter to control the effect of the denoiser and to accelerate the solution. Expand
Async-RED: A Provably Convergent Asynchronous Block Parallel Stochastic Method using Deep Denoising Priors
TLDR
This work proposes a new asynchronous RED (ASYNC-RED) algorithm that enables asynchronous parallel processing of data, making it significantly faster than its serial counterparts for large-scale inverse problems. Expand
A SIMPLE SPARSE DENOISING LAYER FOR ROBUST DEEP LEARNING
  • 2020
Deep models have achieved great success in many applications. However, vanilla deep models are not well-designed against the input perturbation. In this work, we take an initial step to design aExpand
Patient-specific hyperparameter learning for optimization-based CT image reconstruction
  • Jingyan Xu, F. Noo
  • Physics, Medicine
  • Physics in medicine and biology
  • 2021
TLDR
A hyperparameter learning framework that learns patient-specific hyperparameters for optimization based image reconstruction problems for x-ray CT applications with exactly computable solutions so that the whole network can be trained end-to end in an efficient manner. Expand
ASYNC-RED: A PROVABLY CONVERGENT ASYN-
Regularization by denoising (RED) is a recently developed framework for solving inverse problems by integrating advanced denoisers as image priors. Recent work has shown its state-of-the-artExpand
Accelerating Quadratic Optimization with Reinforcement Learning
TLDR
In experiments with well-known QP benchmarks, the RL policy, RLQP, significantly outperforms state-of-the-art QP solvers by up to 3x and generalizes surprisingly well to previously unseen problems with varying dimension and structure. Expand
Compressed Sensing-Based Robust Phase Retrieval via Deep Generative Priors
TLDR
This article proposes a framework to regularize the highly ill-posed and non-linear phase retrieval problem through deep generative priors by simply using gradient descent algorithm and modify the proposed algorithm to allow the generative model to explore solutions outside its range, leading to improved performance. Expand
Dynamic Proximal Unrolling Network for Compressive Sensing Imaging
TLDR
A dynamic proximal unrolling network (dubbed DPUNet), which can handle a variety of measurement matrices via one single model without retraining, and outperform the state-of-the-art approaches. Expand
FISTA-Net: Learning a Fast Iterative Shrinkage Thresholding Network for Inverse Problems in Imaging
TLDR
A model-based deep learning network, named FISTA-Net, is proposed by combining the merits of interpretability and generality of the model- based Fast Iterative Shrinkage/Thresholding Algorithm (FISTA) and strong regularization and tuning-free advantages of the data-driven neural network. Expand
...
1
2
3
...

References

SHOWING 1-10 OF 110 REFERENCES
Parameter-free Plug-and-Play ADMM for image restoration
  • Xiran Wang, S. Chan
  • Mathematics, Computer Science
  • 2017 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)
  • 2017
TLDR
This paper presents a parameter-free Plug-and-Play ADMM where internal parameters are updated as part of the optimization, derived from the generalized approximate message passing, with several essential modifications. Expand
Plug-and-Play Methods Provably Converge with Properly Trained Denoisers
Plug-and-play (PnP) is a non-convex framework that integrates modern denoising priors, such as BM3D or deep learning-based denoisers, into ADMM or other proximal algorithms. An advantage of PnP isExpand
An Online Plug-and-Play Algorithm for Regularized Image Reconstruction
TLDR
A new online PnP algorithm based on the proximal gradient method (PGM) is introduced, which uses only a subset of measurements at every iteration, which makes it scalable to very large datasets. Expand
A Plug-and-Play Priors Approach for Solving Nonlinear Imaging Inverse Problems
TLDR
This letter develops the fast iterative shrinkage/thresholding algorithm variant of PPP for model-based nonlinear inverse scattering and shows that the PPP approach is applicable beyond linear inverse problems. Expand
Optimizing a Parameterized Plug-and-Play ADMM for Iterative Low-Dose CT Reconstruction
  • Ji He, Yan Yang, +6 authors J. Ma
  • Computer Science, Medicine
  • IEEE Transactions on Medical Imaging
  • 2019
TLDR
Experimental results obtained on clinical patient datasets demonstrate that the proposed deep learning-based strategy for MBIR can achieve promising gains over existing algorithms for LdCT image reconstruction in terms of noise-induced artifact suppression and edge detail preservation. Expand
Image Restoration by Iterative Denoising and Backward Projections
TLDR
An alternative method for solving inverse problems using off-the-shelf denoisers, which requires less parameter tuning and is competitiveness with task-specific techniques and the P&P approach for image inpainting and deblurring is proposed. Expand
Primal-Dual Plug-and-Play Image Restoration
  • S. Ono
  • Mathematics, Computer Science
  • IEEE Signal Processing Letters
  • 2017
TLDR
This approach resolves issues by leveraging the nature of primal-dual splitting, yielding a very flexible plug-and-play image restoration method that is much more efficient than ADMMPnP with an inner loop and keeps the same efficiency in the case where the subproblem of ADM MPnP can be solved efficiently. Expand
Plug-and-Play priors for model based reconstruction
TLDR
This paper demonstrates with some simple examples how Plug-and-Play priors can be used to mix and match a wide variety of existing denoising models with a tomographic forward model, thus greatly expanding the range of possible problem solutions. Expand
Learning Proximal Operators: Using Denoising Networks for Regularizing Inverse Imaging Problems
TLDR
This paper studies the possibility of replacing the proximal operator of the regularization used in many convex energy minimization algorithms by a denoising neural network, and obtains state-of-the-art reconstruction results. Expand
Performance Analysis of Plug-and-Play ADMM: A Graph Signal Processing Perspective
  • S. Chan
  • Computer Science
  • IEEE Transactions on Computational Imaging
  • 2019
TLDR
This paper restricts the denoisers to the class of graph filters under a linearity assumption, or more specifically the symmetric smoothing filters, and introduces a new analysis technique via the concept of consensus equilibrium to provide interpretations to problems involving multiple priors. Expand
...
1
2
3
4
5
...