Stochastic Deep Restoration Priors for Imaging Inverse Problems

Yuyang Hu, Albert Peng, Weijie Gan, Peyman Milanfar, Mauricio Delbracio, Ulugbek S. Kamilov
Proceedings of the 42nd International Conference on Machine Learning, PMLR 267:24621-24652, 2025.

Abstract

Deep neural networks trained as image denoisers are widely used as priors for solving imaging inverse problems. We introduce Stochastic deep Restoration Priors (ShaRP), a novel framework that stochastically leverages an ensemble of deep restoration models beyond denoisers to regularize inverse problems. By using generalized restoration models trained on a broad range of degradations beyond simple Gaussian noise, ShaRP effectively addresses structured artifacts and enables self-supervised training without fully sampled data. We prove that ShaRP minimizes an objective function involving a regularizer derived from the score functions of minimum mean square error (MMSE) restoration operators. We also provide theoretical guarantees for learning restoration operators from incomplete measurements. ShaRP achieves state-of-the-art performance on tasks such as magnetic resonance imaging reconstruction and single-image super-resolution, surpassing both denoiser- and diffusion-model-based methods without requiring retraining.

Cite this Paper

BibTeX
@InProceedings{pmlr-v267-hu25s, title = {Stochastic Deep Restoration Priors for Imaging Inverse Problems}, author = {Hu, Yuyang and Peng, Albert and Gan, Weijie and Milanfar, Peyman and Delbracio, Mauricio and Kamilov, Ulugbek S.}, booktitle = {Proceedings of the 42nd International Conference on Machine Learning}, pages = {24621--24652}, year = {2025}, editor = {Singh, Aarti and Fazel, Maryam and Hsu, Daniel and Lacoste-Julien, Simon and Berkenkamp, Felix and Maharaj, Tegan and Wagstaff, Kiri and Zhu, Jerry}, volume = {267}, series = {Proceedings of Machine Learning Research}, month = {13--19 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v267/main/assets/hu25s/hu25s.pdf}, url = {https://proceedings.mlr.press/v267/hu25s.html}, abstract = {Deep neural networks trained as image denoisers are widely used as priors for solving imaging inverse problems. We introduce Stochastic deep Restoration Priors (ShaRP), a novel framework that stochastically leverages an ensemble of deep restoration models beyond denoisers to regularize inverse problems. By using generalized restoration models trained on a broad range of degradations beyond simple Gaussian noise, ShaRP effectively addresses structured artifacts and enables self-supervised training without fully sampled data. We prove that ShaRP minimizes an objective function involving a regularizer derived from the score functions of minimum mean square error (MMSE) restoration operators. We also provide theoretical guarantees for learning restoration operators from incomplete measurements. ShaRP achieves state-of-the-art performance on tasks such as magnetic resonance imaging reconstruction and single-image super-resolution, surpassing both denoiser- and diffusion-model-based methods without requiring retraining.} }
Endnote
%0 Conference Paper %T Stochastic Deep Restoration Priors for Imaging Inverse Problems %A Yuyang Hu %A Albert Peng %A Weijie Gan %A Peyman Milanfar %A Mauricio Delbracio %A Ulugbek S. Kamilov %B Proceedings of the 42nd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2025 %E Aarti Singh %E Maryam Fazel %E Daniel Hsu %E Simon Lacoste-Julien %E Felix Berkenkamp %E Tegan Maharaj %E Kiri Wagstaff %E Jerry Zhu %F pmlr-v267-hu25s %I PMLR %P 24621--24652 %U https://proceedings.mlr.press/v267/hu25s.html %V 267 %X Deep neural networks trained as image denoisers are widely used as priors for solving imaging inverse problems. We introduce Stochastic deep Restoration Priors (ShaRP), a novel framework that stochastically leverages an ensemble of deep restoration models beyond denoisers to regularize inverse problems. By using generalized restoration models trained on a broad range of degradations beyond simple Gaussian noise, ShaRP effectively addresses structured artifacts and enables self-supervised training without fully sampled data. We prove that ShaRP minimizes an objective function involving a regularizer derived from the score functions of minimum mean square error (MMSE) restoration operators. We also provide theoretical guarantees for learning restoration operators from incomplete measurements. ShaRP achieves state-of-the-art performance on tasks such as magnetic resonance imaging reconstruction and single-image super-resolution, surpassing both denoiser- and diffusion-model-based methods without requiring retraining.
APA
Hu, Y., Peng, A., Gan, W., Milanfar, P., Delbracio, M. & Kamilov, U.S.. (2025). Stochastic Deep Restoration Priors for Imaging Inverse Problems. Proceedings of the 42nd International Conference on Machine Learning, in Proceedings of Machine Learning Research 267:24621-24652 Available from https://proceedings.mlr.press/v267/hu25s.html.

Related Material