{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,29]],"date-time":"2026-10-29T12:32:56Z","timestamp":1793277176499,"version":"4.5.0"},"reference-count":52,"publisher":"Society for Industrial & Applied Mathematics (SIAM)","issue":"4","content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["SIAM J. Imaging Sci."],"published-print":{"date-parts":[[2026,12,31]]},"abstract":"<jats:p>Abstract.<\/jats:p>\n                  <jats:p>Many crucial tasks in image processing and computer vision are formulated as inverse problems. Therefore, it is of great importance to design efficient and robust algorithms to solve these problems. In this paper, we focus on generalized projected gradient descent (GPGD) algorithms where generalized projections are realized with learned neural networks, as they provide state-of-the-art results for imaging inverse problems.\u00a0Indeed, neural networks allow for projections onto unknown low-dimensional sets that model complex data, such as images. We call these projections deep projective priors. In generic settings, when the orthogonal projection onto a low-dimensional model set is used, it has been shown that, with a restricted isometry assumption, the corresponding orthogonal PGD converges with a linear rate, yielding near-optimal convergence (within the class of GPGD methods) in the classical case of sparse recovery.\u00a0However, for deep projective priors trained with classical mean squared error losses, there is little guarantee that the hypotheses for linear convergence are satisfied. In this paper, we propose a stochastic orthogonal regularization of the training loss for deep projective priors. This regularization is motivated by our theoretical results:\u00a0a sufficiently good approximation of the orthogonal projection guarantees linear stable recovery with performance close to orthogonal PGD. We show experimentally, using two different deep projective priors (based on autoencoders and on denoising networks), that our stochastic orthogonal regularization yields projections that improve convergence speed and robustness of GPGD in challenging inverse problem settings, in accordance with our theoretical findings.<\/jats:p>","DOI":"10.1137\/25m1824835","type":"journal-article","created":{"date-parts":[[2026,10,29]],"date-time":"2026-10-29T11:47:39Z","timestamp":1793274459000},"page":"2306-2339","source":"Crossref","is-referenced-by-count":0,"title":["Stochastic Orthogonal Regularization for Deep Projective Priors"],"prefix":"10.1137","volume":"19","author":[{"ORCID":"https:\/\/orcid.org\/0009-0004-1451-3808","authenticated-orcid":true,"given":"Ali","family":"Joundi","sequence":"first","affiliation":[{"name":"Corresponding author.\u00a0Universit\u00e9 de Bordeaux, CNRS, Bordeaux INP, IMB, UMR 5251 33400 Talence, France."}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-1826-4760","authenticated-orcid":true,"given":"Yann","family":"Traonmilin","sequence":"additional","affiliation":[{"name":"Universit\u00e9 de Bordeaux, CNRS, Bordeaux INP, IMB, UMR 5251 33400 Talence, France."}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Alasdair","family":"Newson","sequence":"additional","affiliation":[{"name":"ISIR, Sorbonne Universit\u00e9, Paris, France."}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"351","published-online":{"date-parts":[[2026,10,29]]},"reference":[{"key":"ref1","doi-asserted-by":"publisher","DOI":"10.1109\/TIT.2011.2146550"},{"key":"ref2","doi-asserted-by":"publisher","DOI":"10.1109\/LSP.2004.838190"},{"key":"ref3","doi-asserted-by":"publisher","DOI":"10.1109\/TCI.2016.2629286"},{"key":"ref4","unstructured":"H. Chung, J. Kim, M. T. Mccann, M. L. Klasky, and J. C. Ye, Diffusion Posterior Sampling for General Noisy Inverse Problems, preprint, arXiv:2209.14687, 2022."},{"key":"ref5","doi-asserted-by":"publisher","DOI":"10.52202\/068431-1862"},{"key":"ref6","doi-asserted-by":"crossref","unstructured":"H. Chung, B. Sim, and J. C. Ye, Come-closer-diffuse-faster:\u00a0Accelerating conditional diffusion models for inverse problems through stochastic contraction, in Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition, 2022, pp. 12413\u201312422.","DOI":"10.1109\/CVPR52688.2022.01209"},{"key":"ref7","doi-asserted-by":"publisher","DOI":"10.1137\/20M1337168"},{"key":"ref8","unstructured":"G. Daras, H. Chung, C.H. Lai, Y. Mitsufuji, J. C. Ye, P. Milanfar, A. G. Dimakis, and M. Delbracio, A Survey on Diffusion Models for Inverse Problems, preprint, arXiv:2410.00083, 2024."},{"key":"ref9","unstructured":"S. Diamond, V. Sitzmann, F. Heide, and G. Wetzstein, Unrolled Optimization with Deep Priors, preprint, arXiv:1705.08041, 2017."},{"key":"ref10","doi-asserted-by":"crossref","unstructured":"C. Dong, C. C. Loy, K. He, and X. Tang, Learning a deep convolutional network for image super-resolution, in Computer Vision\u2013ECCV 2014:\u00a013th European Conference, Zurich, Switzerland, September 6-12, 2014, Proceedings, Part IV 13, Springer, 2014, pp. 184\u2013199.","DOI":"10.1007\/978-3-319-10593-2_13"},{"key":"ref11","doi-asserted-by":"publisher","DOI":"10.1007\/978-0-8176-4948-7"},{"key":"ref12","doi-asserted-by":"publisher","DOI":"10.1109\/TIT.2018.2841379"},{"key":"ref13","doi-asserted-by":"publisher","DOI":"10.1137\/21M140225X"},{"key":"ref14","doi-asserted-by":"publisher","DOI":"10.1007\/s10994-020-05929-w"},{"key":"ref15","doi-asserted-by":"publisher","DOI":"10.1109\/TMI.2018.2832656"},{"key":"ref16","volume-title":"IEEE Trans. Pattern Anal. Mach. Intell.","author":"He C.","year":"2025"},{"key":"ref17","unstructured":"S. Hurault, A. Leclaire, and N. Papadakis, Proximal denoiser for convergent plug-and-play optimization with nonconvex regularization, in International Conference on Machine Learning, PMLR, 2022, pp. 9483\u20139505."},{"key":"ref18","volume-title":"Adv. Neural Inform. Process. Syst.","volume":"27","author":"Jain P.","year":"2014"},{"key":"ref19","unstructured":"A. Joundi, Y. Traonmilin, and J.F. Aujol, From sparse recovery to plug-and-play priors, understanding trade-offs for stable recovery with generalized projected gradient descent, in Conference on Parsimony and Learning, 2026."},{"key":"ref20","unstructured":"A. Joundi, Y. Traonmilin, and A. Newson, Code to reproduce experiments, released at publication, 2024."},{"key":"ref21","first-page":"13242","volume":"34","author":"Kadkhodaie Z.","year":"2021","journal-title":"Adv. Neural Inform. Process. Syst."},{"key":"ref22","doi-asserted-by":"publisher","DOI":"10.1109\/MSP.2022.3199595"},{"key":"ref23","doi-asserted-by":"publisher","DOI":"10.1109\/LSP.2017.2763583"},{"key":"ref24","doi-asserted-by":"publisher","DOI":"10.52202\/068431-1714"},{"key":"ref25","unstructured":"O. Leong and Y. Traonmilin, A recovery theory for diffusion priors:\u00a0Deterministic analysis of the implicit prior algorithm, in The 29th International Conference on Artificial Intelligence and Statistics, 2026."},{"key":"ref26","unstructured":"X. Li, Z. Shen, Y.P. Hsieh, and N. He, When Scores Learn Geometry:\u00a0Rate Separations under the Manifold Hypothesis, preprint, arXiv:2509.24912, 2025."},{"key":"ref27","first-page":"5921","volume":"34","author":"Liu J.","year":"2021","journal-title":"Adv. Neural Inform. Process. Syst."},{"key":"ref28","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-319-70139-4"},{"key":"ref29","doi-asserted-by":"crossref","unstructured":"Z. Liu, P. Luo, X. Wang, and X. Tang, Deep learning face attributes in the wild, in Proceedings of International Conference on Computer Vision (ICCV), 2015.","DOI":"10.1109\/ICCV.2015.425"},{"key":"ref30","unstructured":"S. Martin, A. Gagneux, P. Hagemann, and G. Steidl, PNP-flow:\u00a0Plug-and-play Image Restoration with Flow Matching, preprint, arXiv:2410.02423, 2024."},{"key":"ref31","doi-asserted-by":"crossref","unstructured":"T. Meinhardt, M. Moller, C. Hazirbas, and D. Cremers, Learning proximal operators:\u00a0Using denoising networks for regularizing inverse imaging problems, in Proceedings of the IEEE International Conference on Computer Vision, 2017, pp. 1781\u20131790.","DOI":"10.1109\/ICCV.2017.198"},{"key":"ref32","unstructured":"T. Miyato, T. Kataoka, M. Koyama, and Y. Yoshida, Spectral Normalization for Generative Adversarial Networks, preprint, arXiv:1802.05957, 2018."},{"key":"ref33","doi-asserted-by":"publisher","DOI":"10.1109\/JSAIT.2020.2991563"},{"key":"ref34","doi-asserted-by":"publisher","DOI":"10.1109\/JSAIT.2020.2983643"},{"key":"ref35","doi-asserted-by":"publisher","DOI":"10.1137\/20M1387961"},{"key":"ref36","doi-asserted-by":"crossref","unstructured":"K. Ranasinghe, M. Naseer, M. Hayat, S. Khan, and F. S. Khan, Orthogonal projection loss, in Proceedings of the IEEE\/CVF International Conference on Computer Vision, 2021, pp. 12333\u201312343.","DOI":"10.1109\/ICCV48922.2021.01211"},{"key":"ref37","unstructured":"M. Renaud, J. Prost, A. Leclaire, and N. Papadakis, Plug-and-play image restoration with stochastic denoising regularization, in Forty-first International Conference on Machine Learning, 2024."},{"key":"ref38","doi-asserted-by":"publisher","DOI":"10.1137\/16M1102884"},{"key":"ref39","unstructured":"E. Ryu, J. Liu, S. Wang, X. Chen, Z. Wang, and W. Yin, Plug-and-play methods provably converge with properly trained denoisers, in International Conference on Machine Learning, PMLR, 2019, pp. 5546\u20135557."},{"key":"ref40","doi-asserted-by":"publisher","DOI":"10.1109\/JSAIT.2023.3241123"},{"key":"ref41","first-page":"4609","volume-title":"2018 IEEE ICASSP, IEEE","author":"Shah V.","year":"2018"},{"key":"ref42","doi-asserted-by":"crossref","unstructured":"R. N. Spreng, R. Setton, U. Alter, B. N. Cassidy, B. Darboh, E. DuPre, K. Kantarovich, A. W. Lockrow, L. Mwilambwe-Tshilobo, W.M. Luh, P. Kundu, and G. R. Turner, neurocognitive aging data release with behavioral, structural, and multi-echo functional MRI measures, 2022, https:\/\/doi.org\/10.1038\/s41597-022-01231-7.","DOI":"10.1038\/s41597-022-01231-7"},{"key":"ref43","doi-asserted-by":"publisher","DOI":"10.1109\/TCI.2019.2893568"},{"key":"ref44","doi-asserted-by":"crossref","unstructured":"Y. Traonmilin, J.F. Aujol, and A. Guennec, Towards Optimal Algorithms for the Recovery of Low-Dimensional Models with Linear Rates, preprint, arXiv:2410.06607, 2024.","DOI":"10.2139\/ssrn.5353067"},{"key":"ref45","doi-asserted-by":"crossref","unstructured":"S. V. Venkatakrishnan, C. A. Bouman, and B. Wohlberg, Plug-and-play priors for model based reconstruction, in 2013 IEEE Global Conference on Signal and Information Processing, IEEE, 2013, pp. 945\u2013948.","DOI":"10.1109\/GlobalSIP.2013.6737048"},{"key":"ref46","doi-asserted-by":"crossref","unstructured":"J. Wang, Y. Chen, R. Chakraborty, and S. X. Yu, Orthogonal convolutional neural networks, in Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition, 2020, pp. 11505\u201311515.","DOI":"10.1109\/CVPR42600.2020.01152"},{"key":"ref47","doi-asserted-by":"publisher","DOI":"10.1109\/ACCESS.2019.2916030"},{"key":"ref48","doi-asserted-by":"publisher","DOI":"10.1109\/TPAMI.2020.2982166"},{"key":"ref49","doi-asserted-by":"publisher","DOI":"10.1016\/j.sigpro.2016.05.002"},{"key":"ref50","doi-asserted-by":"publisher","DOI":"10.1109\/TPAMI.2021.3088914"},{"key":"ref51","doi-asserted-by":"publisher","DOI":"10.1109\/TIP.2018.2839891"},{"key":"ref52","doi-asserted-by":"publisher","DOI":"10.52202\/079017-1829"}],"container-title":["SIAM Journal on Imaging Sciences"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/epubs.siam.org\/doi\/pdf\/10.1137\/25M1824835","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,10,29]],"date-time":"2026-10-29T11:47:48Z","timestamp":1793274468000},"score":1,"resource":{"primary":{"URL":"https:\/\/epubs.siam.org\/doi\/10.1137\/25M1824835"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,10,29]]},"references-count":52,"journal-issue":{"issue":"4","published-print":{"date-parts":[[2026,12,31]]}},"alternative-id":["10.1137\/25M1824835"],"URL":"https:\/\/doi.org\/10.1137\/25m1824835","relation":{},"ISSN":["1936-4954"],"issn-type":[{"value":"1936-4954","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,10,29]]}}}