{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,11,9]],"date-time":"2026-11-09T22:36:15Z","timestamp":1794263775684,"version":"4.9.0"},"reference-count":53,"publisher":"Society for Industrial & Applied Mathematics (SIAM)","issue":"2","funder":[{"DOI":"10.13039\/501100001665","name":"Agence Nationale de la Recherche","doi-asserted-by":"publisher","award":["TASK- ABILE ANR-22-CE48-0010"],"award-info":[{"award-number":["TASK- ABILE ANR-22-CE48-0010"]}],"id":[{"id":"10.13039\/501100001665","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/100010663","name":"H2020 European Research Council","doi-asserted-by":"publisher","award":["101117133"],"award-info":[{"award-number":["101117133"]}],"id":[{"id":"10.13039\/100010663","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100000780","name":"European Commission","doi-asserted-by":"publisher","award":["TraDE-OPT 861137"],"award-info":[{"award-number":["TraDE-OPT 861137"]}],"id":[{"id":"10.13039\/501100000780","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/100010663","name":"H2020 European Research Council","doi-asserted-by":"publisher","award":["SLING 819789"],"award-info":[{"award-number":["SLING 819789"]}],"id":[{"id":"10.13039\/100010663","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/100000181","name":"Air Force Office of Scientific Research","doi-asserted-by":"publisher","award":["FA8655-22-1-7034"],"award-info":[{"award-number":["FA8655-22-1-7034"]}],"id":[{"id":"10.13039\/100000181","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100003407","name":"Ministry of Education, University and Research","doi-asserted-by":"crossref","award":["PRIN 202244A7YL"],"award-info":[{"award-number":["PRIN 202244A7YL"]}],"id":[{"id":"10.13039\/501100003407","id-type":"DOI","asserted-by":"crossref"}]},{"name":"PNRR","award":["PE0000013 - SPOKE 10"],"award-info":[{"award-number":["PE0000013 - SPOKE 10"]}]},{"DOI":"10.13039\/501100003407","name":"Ministero dell\u2019Istruzione, dell\u2019Universit\u00e0 e della Ricerca","doi-asserted-by":"publisher","award":["CUP D33C23001110001"],"award-info":[{"award-number":["CUP D33C23001110001"]}],"id":[{"id":"10.13039\/501100003407","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/100012740","name":"Gruppo Nazionale per l\u2019Analisi Matematica, la Probabilit\u00e0 e le loro Applicazioni","doi-asserted-by":"publisher","id":[{"id":"10.13039\/100012740","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100001665","name":"Agence Nationale de la Recherche","doi-asserted-by":"publisher","award":["ANR-23-PEIA-0004"],"award-info":[{"award-number":["ANR-23-PEIA-0004"]}],"id":[{"id":"10.13039\/501100001665","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100001665","name":"Agence Nationale de la Recherche","doi-asserted-by":"publisher","award":["ANR-23-IACL-0001"],"award-info":[{"award-number":["ANR-23-IACL-0001"]}],"id":[{"id":"10.13039\/501100001665","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100001665","name":"Agence Nationale de la Recherche","doi-asserted-by":"publisher","award":["ANR-24-CE23-1529"],"award-info":[{"award-number":["ANR-24-CE23-1529"]}],"id":[{"id":"10.13039\/501100001665","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["SIAM J. Imaging Sci."],"published-print":{"date-parts":[[2026,6,30]]},"abstract":"<jats:p>Abstract.<\/jats:p>\n                  <jats:p>Deep equilibrium models (DEQs) are implicit neural networks with fixed points that have recently gained attention for learning image regularization functionals, particularly in settings involving Gaussian fidelities, where assumptions on the forward operator ensure contractiveness of standard (proximal) gradient descent operators. In this work, we extend the application of DEQs to Poisson inverse problems, where the data fidelity term is more appropriately modeled by the Kullback\u2013Leibler divergence. To this end, we introduce a novel DEQ formulation based on mirror descent defined in terms of a tailored non-Euclidean geometry that naturally adapts with the structure of the data term. This enables the learning of neural regularizers within a principled training framework. We derive sufficient conditions and establish refined convergence results based on the Kurdyka\u2013\u0141ojasiewicz framework for functions with nonclosed domains to guarantee the convergence of the learned reconstruction scheme and propose computational strategies that enable both efficient training and parameter-free inference. Numerical experiments show that our method outperforms traditional model-based approaches, and it is comparable to the performance of Bregman plug-and-play methods, while mitigating their typical drawbacks, such as time-consuming tuning of hyperparameters. The code is publicly available at https:\/\/github.com\/christiandaniele\/DEQ-MD .<\/jats:p>","DOI":"10.1137\/25m1779048","type":"journal-article","created":{"date-parts":[[2026,5,12]],"date-time":"2026-05-12T07:13:02Z","timestamp":1778569982000},"page":"1077-1109","source":"Crossref","is-referenced-by-count":2,"title":["Deep Equilibrium Models for Poisson Imaging Inverse Problems via Mirror Descent"],"prefix":"10.1137","volume":"19","author":[{"given":"Christian","family":"Daniele","sequence":"first","affiliation":[{"name":"Computer Science, MaLGa Center, DIBRIS, University of Genoa, Genoa, Italy."}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-6232-5631","authenticated-orcid":true,"given":"Silvia","family":"Villa","sequence":"additional","affiliation":[{"name":"MaLGa Center, DIMA, University of Genoa, Genoa, Italy."}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-4077-708X","authenticated-orcid":true,"given":"Samuel","family":"Vaiter","sequence":"additional","affiliation":[{"name":"CNRS & Universit\u00e9 C\u00f4te d\u2019Azur, Laboratoire J. A. Dieudonn\u00e9, Nice, France."}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Luca","family":"Calatroni","sequence":"additional","affiliation":[{"name":"MaLGa Center, DIBRIS, University of Genoa & MMS, Istituto Italiano di Tecnologia (IIT), Italy."}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"351","published-online":{"date-parts":[[2026,5,12]]},"reference":[{"key":"ref1","doi-asserted-by":"publisher","DOI":"10.1109\/TMI.2018.2799231"},{"key":"ref2","doi-asserted-by":"crossref","unstructured":"A. H. Al-Shabili, X. Xu, I. Selesnick, and U. S. Kamilov, Bregman plug-and-play priors, in 2022 IEEE International Conference on Image Processing (ICIP), IEEE, 2022, pp. 241\u2013245.","DOI":"10.1109\/ICIP46576.2022.9897933"},{"key":"ref3","first-page":"690","volume":"32","author":"Bai S.","year":"2019","journal-title":"Advances in Neural Information Processing Systems"},{"key":"ref4","doi-asserted-by":"publisher","DOI":"10.1287\/moor.2016.0817"},{"key":"ref5","doi-asserted-by":"publisher","DOI":"10.1088\/0266-5611\/25\/12\/123006"},{"key":"ref6","doi-asserted-by":"publisher","DOI":"10.1137\/050644641"},{"key":"ref7","first-page":"13537","volume-title":"Advances in Neural Information Processing Systems 34","author":"Bolte J.","year":"2021"},{"key":"ref8","doi-asserted-by":"publisher","DOI":"10.52202\/068431-1915"},{"key":"ref9","doi-asserted-by":"publisher","DOI":"10.52202\/075280-3371"},{"key":"ref10","doi-asserted-by":"publisher","DOI":"10.1137\/17M1138558"},{"key":"ref11","doi-asserted-by":"publisher","DOI":"10.1137\/17M116001X"},{"key":"ref12","doi-asserted-by":"publisher","DOI":"10.1088\/0266-5611\/25\/1\/015002"},{"key":"ref13","doi-asserted-by":"publisher","DOI":"10.1109\/83.650854"},{"key":"ref14","doi-asserted-by":"publisher","DOI":"10.1109\/83.210869"},{"key":"ref15","doi-asserted-by":"publisher","DOI":"10.1137\/17M1134834"},{"key":"ref16","doi-asserted-by":"publisher","DOI":"10.3390\/jimaging7060099"},{"key":"ref17","doi-asserted-by":"publisher","DOI":"10.1109\/TIP.2008.2008223"},{"key":"ref18","unstructured":"Z. Fang, S. Buchanan, and J. Sulam, What\u2019s in a prior? Learned proximal networks for inverse problems, in International Conference on Learning Representations, 2024, 15652."},{"key":"ref19","doi-asserted-by":"publisher","DOI":"10.1109\/TIP.2010.2053941"},{"key":"ref20","doi-asserted-by":"crossref","unstructured":"S. W. Fung, H. Heaton, Q. Li, D. McKenzie, S. Osher, and W. Yin, JFB: Jacobian-free backpropagation for implicit networks, in Proceedings of the AAAI Conference on Artificial Intelligence, Vol. 36, 2022, pp. 6648\u20136656.","DOI":"10.1609\/aaai.v36i6.20619"},{"key":"ref21","doi-asserted-by":"publisher","DOI":"10.1109\/TCI.2021.3118944"},{"key":"ref22","doi-asserted-by":"crossref","unstructured":"J.J. Godeme, J. Fadili, C. Amra, and M. Zerrad, Stable Phase Retrieval with Mirror Descent, preprint, arXiv:2405.10754, 2024.","DOI":"10.1137\/22M1528896"},{"key":"ref23","doi-asserted-by":"publisher","DOI":"10.1109\/TIP.2011.2168410"},{"key":"ref24","doi-asserted-by":"publisher","DOI":"10.52202\/075280-1187"},{"key":"ref25","unstructured":"S. Hurault, A. Leclaire, and N. Papadakis, Gradient step denoiser for convergent plug-and-play, in International Conference on Learning Representations, 2021, https:\/\/openreview.net\/forum?id=fPhKeld3Okz."},{"key":"ref26","doi-asserted-by":"publisher","DOI":"10.1109\/MSP.2022.3199595"},{"key":"ref27","unstructured":"D. P. Kingma and J. Ba, Adam: A method for stochastic optimization, in International Conference on Learning Representations (ICLR), 2015, https:\/\/arxiv.org\/abs\/1412.6980."},{"key":"ref28","doi-asserted-by":"crossref","unstructured":"T. Klatzer, S. Melidonis, M. Pereyra, and K. C. Zygalakis, Efficient Bayesian Computation Using Plug-and-Play Priors for Poisson Inverse Problems, preprint, arXiv:2503.16222, 2025.","DOI":"10.1137\/25M1736062"},{"key":"ref29","doi-asserted-by":"publisher","DOI":"10.1137\/21M1406349"},{"key":"ref30","doi-asserted-by":"crossref","unstructured":"A. Levin, Y. Weiss, F. Durand, and W. T. Freeman, Understanding and evaluating blind deconvolution algorithms, in 2009 IEEE Conference on Computer Vision and Pattern Recognition, IEEE, 2009, pp. 1964\u20131971.","DOI":"10.1109\/CVPR.2009.5206815"},{"key":"ref31","doi-asserted-by":"publisher","DOI":"10.1137\/16M1099546"},{"key":"ref32","doi-asserted-by":"publisher","DOI":"10.1086\/111605"},{"key":"ref33","doi-asserted-by":"crossref","unstructured":"T. Meinhardt, M. Moller, C. Hazirbas, and D. Cremers, Learning proximal operators: Using 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":"ref34","doi-asserted-by":"publisher","DOI":"10.1109\/MSP.2020.3016905"},{"key":"ref35","series-title":"Wiley-Intersci. Ser. Discrete Math.","volume-title":"Problem Complexity and Method Efficiency in Optimization","author":"Nemirovsky A. S.","year":"1983"},{"key":"ref36","doi-asserted-by":"publisher","DOI":"10.1137\/21M1391699"},{"key":"ref37","doi-asserted-by":"publisher","DOI":"10.1364\/JOSA.62.000055"},{"key":"ref38","series-title":"Princeton Landmarks Math. 28","volume-title":"Convex Analysis","author":"Rockafellar R. T.","year":"1997"},{"key":"ref39","doi-asserted-by":"publisher","DOI":"10.1137\/16M1102884"},{"key":"ref40","doi-asserted-by":"publisher","DOI":"10.1016\/0167-2789(92)90242-F"},{"key":"ref41","unstructured":"T. Salimans and J. Ho, Should EBMs model the energy or the score? in Energy Based Models Workshop-ICLR 2021, 2021."},{"key":"ref42","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-030-86664-8_4"},{"key":"ref43","doi-asserted-by":"crossref","unstructured":"A. Sawatzky, C. Brune, J. M\u00fcller, and M. Burger, Total variation processing of images with Poisson statistics, in International Conference on Computer Analysis of Images and Patterns, Springer, 2009, pp. 533\u2013540.","DOI":"10.1007\/978-3-642-03767-2_65"},{"key":"ref44","doi-asserted-by":"publisher","DOI":"10.1109\/TMI.1982.4307558"},{"key":"ref45","doi-asserted-by":"publisher","DOI":"10.1109\/TCI.2019.2893568"},{"key":"ref46","doi-asserted-by":"publisher","DOI":"10.1109\/TCI.2021.3094062"},{"key":"ref47","doi-asserted-by":"publisher","DOI":"10.1007\/s10107-018-1284-2"},{"key":"ref48","doi-asserted-by":"crossref","unstructured":"M. Terris, A. Repetti, J.C. Pesquet, and Y. Wiaux, Building firmly nonexpansive convolutional neural networks, in ICASSP\u00a02020, 2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), IEEE, 2020, pp. 8658\u20138662.","DOI":"10.1109\/ICASSP40776.2020.9054731"},{"key":"ref49","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":"ref50","doi-asserted-by":"publisher","DOI":"10.1137\/10078356X"},{"key":"ref51","doi-asserted-by":"publisher","DOI":"10.1109\/TIP.2017.2662206"},{"key":"ref52","doi-asserted-by":"publisher","DOI":"10.1109\/OJSP.2023.3296036"},{"key":"ref53","doi-asserted-by":"publisher","DOI":"10.1088\/1361-6420\/accdc5"}],"container-title":["SIAM Journal on Imaging Sciences"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/epubs.siam.org\/doi\/pdf\/10.1137\/25M1779048","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,8,21]],"date-time":"2026-08-21T14:06:21Z","timestamp":1787321181000},"score":1,"resource":{"primary":{"URL":"https:\/\/epubs.siam.org\/doi\/10.1137\/25M1779048"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,5,12]]},"references-count":53,"journal-issue":{"issue":"2","published-print":{"date-parts":[[2026,6,30]]}},"alternative-id":["10.1137\/25M1779048"],"URL":"https:\/\/doi.org\/10.1137\/25m1779048","relation":{},"ISSN":["1936-4954"],"issn-type":[{"value":"1936-4954","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,5,12]]}}}