{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,2,28]],"date-time":"2026-02-28T03:25:19Z","timestamp":1772249119816,"version":"3.50.1"},"reference-count":92,"publisher":"Society of Exploration Geophysicists","issue":"3","funder":[{"name":"ML4Seismic Center"},{"DOI":"10.13039\/100008065","name":"Georgia Research Alliance","doi-asserted-by":"crossref","id":[{"id":"10.13039\/100008065","id-type":"DOI","asserted-by":"crossref"}]},{"name":"ML4Seismic Center"},{"DOI":"10.13039\/100008065","name":"Georgia Research Alliance","doi-asserted-by":"crossref","id":[{"id":"10.13039\/100008065","id-type":"DOI","asserted-by":"crossref"}]}],"content-domain":{"domain":["library.seg.org"],"crossmark-restriction":true},"short-container-title":[],"published-print":{"date-parts":[[2023,5,1]]},"abstract":"<jats:title>ABSTRACT<\/jats:title>\n                  <jats:p>Bayesian inference for high-dimensional inverse problems is computationally costly and requires selecting a suitable prior distribution. Amortized variational inference addresses these challenges by pretraining a neural network that acts as a surrogate conditional distribution that approximates the posterior distribution not only for one instance of the observed data but also for the distribution of the data pertaining to a specific inverse problem. When fed previously unseen data, the neural network \u2014 in our case, a conditional normalizing flow \u2014 provides the posterior samples at virtually no cost. However, the accuracy of amortized variational inference relies on the availability of high-fidelity training data, which seldom exist in geophysical inverse problems because of the earth\u2019s heterogeneous subsurface. In addition, the network is prone to errors if evaluated over data that are not drawn from the training data distribution. As such, we have aimed to increase the resilience of amortized variational inference in the presence of moderate data distribution shifts. We achieve this via a correction to the conditional normalizing flow\u2019s latent distribution that improves the approximation to the posterior distribution for the data at hand. The correction involves relaxing the standard Gaussian assumption on the latent distribution and parameterizing it via a Gaussian distribution with an unknown mean and (diagonal) covariance. These unknowns are then estimated by minimizing the Kullback-Leibler divergence between the corrected and the (physics-based) true posterior distributions. Although generic and applicable to other inverse problems by means of a linearized seismic imaging example, we find that our correction step improves the robustness of amortized variational inference with respect to changes in the number of seismic sources, noise variance, and shifts in the prior distribution. This approach, given noisy seismic data simulated via the linearized Born modeling, provides a seismic image with limited artifacts and an assessment of its uncertainty at approximately the same cost as five reverse time migrations.<\/jats:p>","DOI":"10.1190\/geo2022-0472.1","type":"journal-article","created":{"date-parts":[[2023,2,8]],"date-time":"2023-02-08T09:14:06Z","timestamp":1675847646000},"page":"R297-R322","update-policy":"https:\/\/doi.org\/10.1190\/crossmark-policy","source":"Crossref","is-referenced-by-count":21,"title":["Reliable amortized variational inference with physics-based latent distribution correction"],"prefix":"10.1190","volume":"88","author":[{"ORCID":"https:\/\/orcid.org\/0000-0001-8779-2247","authenticated-orcid":false,"given":"Ali","family":"Siahkoohi","sequence":"first","affiliation":[{"name":"Georgia Institute of Technology 1 , School of Computational Science and Engineering, Atlanta, Georgia, USA. alisk@gatech.edu (corresponding author)"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-7777-6622","authenticated-orcid":false,"given":"Gabrio","family":"Rizzuti","sequence":"additional","affiliation":[{"name":"Utrecht University 2 , Department of Mathematics, Utrecht, The Netherlands. g.rizzuti@uu.nl"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-0917-2442","authenticated-orcid":false,"given":"Rafael","family":"Orozco","sequence":"additional","affiliation":[{"name":"Georgia Institute of Technology 1 , School of Computational Science and Engineering, Atlanta, Georgia, USA. rorozco@gatech.edu"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-1180-2167","authenticated-orcid":false,"given":"Felix J.","family":"Herrmann","sequence":"additional","affiliation":[{"name":"Georgia Institute of Technology 1 , School of Computational Science and Engineering, Atlanta, Georgia, USA. felix.herrmann@gatech.edu"}]}],"member":"186","published-online":{"date-parts":[[2023,4,13]]},"reference":[{"key":"2025121212115438900_r1","doi-asserted-by":"crossref","DOI":"10.1088\/1361-6420\/ac28ec","article-title":"Task adapted reconstruction for inverse problems","volume":"38","author":"Adler","year":"2022","journal-title":"Inverse Problems","ISSN":"https:\/\/id.crossref.org\/issn\/0266-5611","issn-type":"print"},{"key":"2025121212115438900_r2","article-title":"Deep Bayesian inversion","author":"Adler","year":"2018"},{"key":"2025121212115438900_r200","author":"Ali Siahkoohi Institution"},{"key":"2025121212115438900_r3","doi-asserted-by":"crossref","unstructured":"Andrle\n              A.\n            \n            \n              Farchmin\n              N.\n            \n            \n              Hagemann\n              P.\n            \n            \n              Heidenreich\n              S.\n            \n            \n              Soltwisch\n              V.\n            \n            \n              Steidl\n              G.\n            \n          , 2021, Invertible neural networks versus MCMC for posterior reconstruction in grazing incidence X-ray fluorescence: International Conference on Scale Space and Variational Methods in Computer Vision, Springer, 528\u2013539.","DOI":"10.1007\/978-3-030-75549-2_42"},{"key":"2025121212115438900_r4","unstructured":"Asim\n              M.\n            \n            \n              Daniels\n              M.\n            \n            \n              Leong\n              O.\n            \n            \n              Ahmed\n              A.\n            \n            \n              Hand\n              P.\n            \n          , 2020, Invertible generative models for inverse problems: Mitigating representation error and dataset bias: Proceedings of the 37th International Conference on Machine Learning, 399\u2013409."},{"key":"2025121212115438900_r5","volume-title":"Parameter estimation and inverse problems","author":"Aster","year":"2018"},{"key":"2025121212115438900_r6","article-title":"An adaptive transport framework for joint and conditional density estimation","author":"Baptista","year":"2020"},{"key":"2025121212115438900_r7","volume-title":"Pattern recognition and machine learning","author":"Bishop","year":"2006"},{"key":"2025121212115438900_r8","doi-asserted-by":"crossref","first-page":"859","DOI":"10.1080\/01621459.2017.1285773","article-title":"Variational inference: A review for statisticians","volume":"112","author":"Blei","year":"2017","journal-title":"Journal of the American Statistical Association"},{"key":"2025121212115438900_r9","unstructured":"Bora\n              A.\n            \n            \n              Jalal\n              A.\n            \n            \n              Price\n              E.\n            \n            \n              Dimakis\n              A. G.\n            \n          , 2017, Compressed sensing using generative models: Proceedings of the 34th International Conference on Machine Learning, 537\u2013546."},{"key":"2025121212115438900_r10","doi-asserted-by":"crossref","first-page":"30055","DOI":"10.1073\/pnas.1912789117","article-title":"The frontier of simulation-based inference","volume":"117","author":"Cranmer","year":"2020","journal-title":"Proceedings of the National Academy of Sciences"},{"key":"2025121212115438900_r11","doi-asserted-by":"crossref","first-page":"372","DOI":"10.1190\/1.1444928","article-title":"Prior information, sampling distributions, and the curse of dimensionality","volume":"66","author":"Curtis","year":"2001","journal-title":"Geophysics","ISSN":"https:\/\/id.crossref.org\/issn\/0016-8033","issn-type":"print"},{"key":"2025121212115438900_r12","unstructured":"Dinh\n              L.\n            \n            \n              Sohl-Dickstein\n              J.\n            \n            \n              Bengio\n              S.\n            \n          , 2016, Density estimation using real NVP: International Conference on Learning Representations."},{"issue":"2","key":"2025121212115438900_r13","doi-asserted-by":"crossref","first-page":"R63","DOI":"10.1190\/geo2017-0321.1","article-title":"Assessing uncertainties in velocity models and images with a fast nonlinear uncertainty quantification method","volume":"83","author":"Ely","year":"2018","journal-title":"Geophysics","ISSN":"https:\/\/id.crossref.org\/issn\/0016-8033","issn-type":"print"},{"issue":"6","key":"2025121212115438900_r14","doi-asserted-by":"crossref","first-page":"R629","DOI":"10.1190\/geo2017-0824.1","article-title":"Uncertainty quantification for inverse problems with weak partial-differential-equation constraints","volume":"83","author":"Fang","year":"2018","journal-title":"Geophysics","ISSN":"https:\/\/id.crossref.org\/issn\/0016-8033","issn-type":"print"},{"key":"2025121212115438900_r15","doi-asserted-by":"crossref","DOI":"10.1201\/b16018","volume-title":"Bayesian data analysis","author":"Gelman","year":"2013"},{"key":"2025121212115438900_r16","doi-asserted-by":"crossref","first-page":"2812","DOI":"10.1063\/1.324142","article-title":"The Born approximation in the theory of the scattering of elastic waves by flaws","volume":"48","author":"Gubernatis","year":"1977","journal-title":"Journal of Applied Physics","ISSN":"https:\/\/id.crossref.org\/issn\/0021-8979","issn-type":"print"},{"key":"2025121212115438900_r17","unstructured":"Herrmann\n              F. J.\n            \n            \n              Siahkoohi\n              A.\n            \n            \n              Rizzuti\n              G.\n            \n          , 2019, Learned imaging with constraints and uncertainty quantification: Neural Information Processing Systems 2019 Deep Inverse Workshop."},{"key":"2025121212115438900_r18","article-title":"Universal approximation property of invertible neural networks","author":"Ishikawa","year":"2022"},{"key":"2025121212115438900_r19","unstructured":"Jin\n              P.\n            \n            \n              Zhang\n              X.\n            \n            \n              Chen\n              Y.\n            \n            \n              Huang\n              S. X.\n            \n            \n              Liu\n              Z.\n            \n            \n              Lin\n              Y.\n            \n          , 2022, Unsupervised learning of full-waveform inversion: Connecting CNN and partial differential equation in a loop: International Conference on Learning Representations."},{"key":"2025121212115438900_r20","doi-asserted-by":"crossref","first-page":"183","DOI":"10.1023\/A:1007665907178","article-title":"An introduction to variational methods for graphical models","volume":"37","author":"Jordan","year":"1999","journal-title":"Machine Learning","ISSN":"https:\/\/id.crossref.org\/issn\/0885-6125","issn-type":"print"},{"key":"2025121212115438900_r21","article-title":"Conditional injective flows for Bayesian imaging","author":"Khorashadizadeh","year":"2022"},{"key":"2025121212115438900_r22","unstructured":"Kim\n              Y.\n            \n            \n              Wiseman\n              S.\n            \n            \n              Miller\n              A.\n            \n            \n              Sontag\n              D.\n            \n            \n              Rush\n              A.\n            \n          , 2018, Semi-amortized variational autoencoders: Proceedings of the 35th International Conference on Machine Learning, 2678\u20132687."},{"key":"2025121212115438900_r23","article-title":"Adam: A method for stochastic optimization","author":"Kingma","year":"2014"},{"key":"2025121212115438900_r24","unstructured":"Kingma\n              D. P.\n            \n            \n              Welling\n              M.\n            \n          , 2014, Auto-encoding variational Bayes: Proceedings of the 2nd International Conference on Learning Representations."},{"key":"2025121212115438900_r25","first-page":"1269","volume-title":"Proceedings of the thirty-seventh conference on uncertainty in artificial intelligence","author":"Kothari","year":"2021"},{"key":"2025121212115438900_r26","doi-asserted-by":"crossref","first-page":"1245","DOI":"10.1093\/gji\/ggaa245","article-title":"Uncertainty quantification in time-lapse seismic imaging: A full-waveform approach","volume":"222","author":"Kotsi","year":"2020","journal-title":"Geophysical Journal International","ISSN":"https:\/\/id.crossref.org\/issn\/0956-540X","issn-type":"print"},{"key":"2025121212115438900_r27","article-title":"Conditional sampling with monotone GANs","author":"Kovachki","year":"2020"},{"key":"2025121212115438900_r28","doi-asserted-by":"crossref","unstructured":"Kruse\n              J.\n            \n            \n              Detommaso\n              G.\n            \n            \n              Scheichl\n              R.\n            \n            \n              K\u00f6the\n              U.\n            \n          , 2021, HINT: Hierarchical invertible neural transport for density estimation and Bayesian inference: Proceedings of the AAAI Conference on Artificial Intelligence.","DOI":"10.1609\/aaai.v35i9.16997"},{"key":"2025121212115438900_r29","doi-asserted-by":"crossref","first-page":"1138","DOI":"10.1190\/1.1443328","article-title":"Iterative asymptotic inversion in the acoustic approximation","volume":"57","author":"Lambar\u00e9","year":"1992","journal-title":"Geophysics","ISSN":"https:\/\/id.crossref.org\/issn\/0016-8033","issn-type":"print"},{"key":"2025121212115438900_r30","article-title":"Simulation intelligence: Towards a new generation of scientific methods","author":"Lavin","year":"2021"},{"key":"2025121212115438900_r31","doi-asserted-by":"crossref","unstructured":"Li\n              C.\n            \n            \n              Chen\n              C.\n            \n            \n              Carlson\n              D.\n            \n            \n              Carin\n              L.\n            \n          , 2016, Preconditioned stochastic gradient Langevin dynamics for deep neural networks: Proceedings of the Thirtieth AAAI Conference on Artificial Intelligence, AAAI Press, 1788\u20131794.","DOI":"10.1609\/aaai.v30i1.10200"},{"key":"2025121212115438900_r32","article-title":"Differentiable Gaussianization layers for inverse problems regularized by deep generative models","author":"Li","year":"2022"},{"key":"2025121212115438900_r33","doi-asserted-by":"crossref","unstructured":"Li\n              D.\n            \n            \n              Denli\n              H.\n            \n            \n              MacDonald\n              C.\n            \n            \n              Basler-Reeder\n              K.\n            \n            \n              Baumstein\n              A.\n            \n            \n              Daves\n              J.\n            \n          , 2021a, Multiparameter geophysical reservoir characterization augmented by generative networks: First International Meeting for Applied Geoscience & Energy, SEG, Expanded Abstracts, 1364\u20131368, doi: http:\/\/dx.doi.org\/10.1190\/segam2021-3583673.1.","DOI":"10.1190\/segam2021-3583673.1"},{"key":"2025121212115438900_r34","unstructured":"Li\n              Z.\n            \n            \n              Kovachki\n              N. B.\n            \n            \n              Azizzadenesheli\n              K.\n            \n            \n              Liu\n              B.\n            \n            \n              Bhattacharya\n              K.\n            \n            \n              Stuart\n              A. M.\n            \n            \n              Anandkumar\n              A.\n            \n          , 2021b, Fourier neural operator for parametric partial differential equations: Proceedings of the 9th International Conference on Learning Representations, OpenReview.net."},{"key":"2025121212115438900_r35","unstructured":"Liu\n              Q.\n            \n            \n              Wang\n              D.\n            \n          , 2016, Stein variational gradient descent: A general purpose Bayesian inference algorithm: Advances in Neural Information Processing Systems, Curran Associates, Inc., 2378\u20132386."},{"key":"2025121212115438900_r36","doi-asserted-by":"crossref","first-page":"331","DOI":"10.1214\/14-BA862","article-title":"On divergence measures leading to Jeffreys and other reference priors","volume":"9","author":"Liu","year":"2014","journal-title":"Bayesian Analysis"},{"key":"2025121212115438900_r37","doi-asserted-by":"crossref","first-page":"1165","DOI":"10.5194\/gmd-12-1165-2019","article-title":"Devito (v3.1.0): An embedded domain-specific language for finite differences and geophysical exploration","volume":"12","author":"Louboutin","year":"2019","journal-title":"Geoscientific Model Development"},{"key":"2025121212115438900_r38","article-title":"Architecture and performance of Devito, a system for automated stencil computation","author":"Luporini","year":"2018"},{"key":"2025121212115438900_r39","doi-asserted-by":"crossref","first-page":"1005","DOI":"10.1190\/1.1778243","article-title":"Expanded uncertainty quantification in inverse problems: Hierarchical Bayes and empirical Bayes","volume":"69","author":"Malinverno","year":"2004","journal-title":"Geophysics","ISSN":"https:\/\/id.crossref.org\/issn\/0016-8033","issn-type":"print"},{"issue":"3","key":"2025121212115438900_r40","doi-asserted-by":"crossref","first-page":"W15","DOI":"10.1190\/1.2194516","article-title":"Two ways to quantify uncertainty in geophysical inverse problems","volume":"71","author":"Malinverno","year":"2006","journal-title":"Geophysics","ISSN":"https:\/\/id.crossref.org\/issn\/0016-8033","issn-type":"print"},{"key":"2025121212115438900_r41","doi-asserted-by":"crossref","first-page":"A1460","DOI":"10.1137\/110845598","article-title":"A stochastic Newton MCMC method for large-scale statistical inverse problems with application to seismic inversion","volume":"34","author":"Martin","year":"2012","journal-title":"SIAM Journal on Scientific Computing","ISSN":"https:\/\/id.crossref.org\/issn\/1064-8275","issn-type":"print"},{"key":"2025121212115438900_r42","doi-asserted-by":"crossref","first-page":"208","DOI":"10.1190\/1.1444517","article-title":"Least-squares migration of incomplete reflection data","volume":"64","author":"Nemeth","year":"1999","journal-title":"Geophysics","ISSN":"https:\/\/id.crossref.org\/issn\/0016-8033","issn-type":"print"},{"key":"2025121212115438900_r43","doi-asserted-by":"crossref","first-page":"1574","DOI":"10.1137\/070704277","article-title":"Robust stochastic approximation approach to stochastic programming","volume":"19","author":"Nemirovski","year":"2009","journal-title":"SIAM Journal on Optimization","ISSN":"https:\/\/id.crossref.org\/issn\/1095-7189","issn-type":"print"},{"key":"2025121212115438900_r44","unstructured":"Orozco\n              R.\n            \n            \n              Siahkoohi\n              A.\n            \n            \n              Rizzuti\n              G.\n            \n            \n              van Leeuwen\n              T.\n            \n            \n              Herrmann\n              F. J.\n            \n          , 2021, Photoacoustic imaging with conditional priors from normalizing flows: NeurIPS 2021 Workshop on Deep Learning and Inverse Problems."},{"key":"2025121212115438900_r45","doi-asserted-by":"crossref","unstructured":"Orozco\n              R.\n            \n            \n              Siahkoohi\n              A.\n            \n            \n              Rizzuti\n              G.\n            \n            \n              van Leeuwen\n              T.\n            \n            \n              Herrmann\n              F. J.\n            \n          , 2023, Adjoint operators enable fast and amortized machine learning based Bayesian uncertainty quantification: SPIE Medical Imaging Conference, SPIE.","DOI":"10.1117\/12.2651691"},{"key":"2025121212115438900_r46","doi-asserted-by":"crossref","first-page":"1114","DOI":"10.1111\/1365-2478.12058","article-title":"Model-uncertainty quantification in seismic tomography: Method and applications","volume":"61","author":"Osypov","year":"2013","journal-title":"Geophysical Prospecting","ISSN":"https:\/\/id.crossref.org\/issn\/0016-8025","issn-type":"print"},{"key":"2025121212115438900_r470","first-page":"1","article-title":"Normalizing flows for probabilistic modeling and inference","volume":"22","author":"Papamakarios","year":"2021","journal-title":"Journal of Machine Learning Research"},{"key":"2025121212115438900_r47","doi-asserted-by":"crossref","first-page":"645","DOI":"10.1137\/17M1134640","article-title":"Transport map accelerated Markov chain Monte Carlo","volume":"6","author":"Parno","year":"2018","journal-title":"SIAM\/ASA Journal on Uncertainty Quantification"},{"key":"2025121212115438900_r48","doi-asserted-by":"crossref","first-page":"2321","DOI":"10.1007\/s10444-019-09711-y","article-title":"A transport-based multifidelity preconditioner for Markov chain Monte Carlo","volume":"45","author":"Peherstorfer","year":"2019","journal-title":"Advances in Computational Mathematics","ISSN":"https:\/\/id.crossref.org\/issn\/1019-7168","issn-type":"print"},{"key":"2025121212115438900_r49","doi-asserted-by":"crossref","first-page":"1452","DOI":"10.1109\/TNNLS.2020.3042395","article-title":"BayesFlow: Learning complex stochastic models with invertible neural networks","volume":"33","author":"Radev","year":"2022","journal-title":"IEEE Transactions on Neural Networks and Learning Systems"},{"key":"2025121212115438900_r50","doi-asserted-by":"crossref","first-page":"522","DOI":"10.1093\/gji\/ggx428","article-title":"Low frequency full waveform seismic inversion within a tree based Bayesian framework","volume":"212","author":"Ray","year":"2017","journal-title":"Geophysical Journal International","ISSN":"https:\/\/id.crossref.org\/issn\/0956-540X","issn-type":"print"},{"key":"2025121212115438900_r51","unstructured":"Ren\n              Y.\n            \n            \n              Witte\n              P. A.\n            \n            \n              Siahkoohi\n              A.\n            \n            \n              Louboutin\n              M.\n            \n            \n              Herrmann\n              F. J.\n            \n          , 2021, Seismic velocity inversion and uncertainty quantification using conditional normalizing flows: AGU Fall Meeting."},{"key":"2025121212115438900_r52","unstructured":"Rezende\n              D.\n            \n            \n              Mohamed\n              S.\n            \n          , 2015, Variational inference with normalizing flows: Proceedings of the 32nd International Conference on Machine Learning, 1530\u20131538."},{"key":"2025121212115438900_r53","doi-asserted-by":"crossref","unstructured":"Rizzuti\n              G.\n            \n            \n              Siahkoohi\n              A.\n            \n            \n              Witte\n              P. A.\n            \n            \n              Herrmann\n              F. J.\n            \n          , 2020, Parameterizing uncertainty by deep invertible networks: An application to reservoir characterization: 90th Annual International Meeting, SEG, Expanded Abstracts, 1541\u20131545, doi: http:\/\/dx.doi.org\/10.1190\/segam2020-3428150.1.","DOI":"10.1190\/segam2020-3428150.1"},{"key":"2025121212115438900_r54","doi-asserted-by":"crossref","first-page":"400","DOI":"10.1214\/aoms\/1177729586","article-title":"A stochastic approximation method","volume":"22","author":"Robbins","year":"1951","journal-title":"The Annals of Mathematical Statistics"},{"key":"2025121212115438900_r55","doi-asserted-by":"crossref","DOI":"10.1007\/978-1-4757-4145-2","volume-title":"Monte Carlo statistical methods","author":"Robert","year":"2004"},{"key":"2025121212115438900_r56","article-title":"Detecting model misspecification in amortized Bayesian inference with neural networks","author":"Schmitt","year":"2021"},{"key":"2025121212115438900_r57","doi-asserted-by":"crossref","unstructured":"Schuster\n              G. T.\n            \n          , 1993, Least-squares cross-well migration: 63rd Annual International Meeting, SEG, Expanded Abstracts, 110\u2013113, doi: http:\/\/dx.doi.org\/10.1190\/1.1822308.","DOI":"10.1190\/1.1822308"},{"key":"2025121212115438900_r58","doi-asserted-by":"crossref","unstructured":"Siahkoohi\n              A.\n            \n            \n              Herrmann\n              F. J.\n            \n          , 2021, Learning by example: Fast reliability-aware seismic imaging with normalizing flows: First International Meeting for Applied Geoscience & Energy, SEG, Expanded Abstracts, 1580\u20131585, doi: http:\/\/dx.doi.org\/10.1190\/segam2021-3581836.1.","DOI":"10.1190\/segam2021-3581836.1"},{"key":"2025121212115438900_r59","doi-asserted-by":"crossref","unstructured":"Siahkoohi\n              A.\n            \n            \n              Louboutin\n              M.\n            \n            \n              Herrmann\n              F. J.\n            \n          , 2022a, Velocity continuation with Fourier neural operators for accelerated uncertainty quantification: 2nd International Meeting for Applied Geoscience & Energy, SEG, 1765\u20131769, doi: http:\/\/dx.doi.org\/10.1190\/image2022-3750475.1.","DOI":"10.1190\/image2022-3750475.1"},{"key":"2025121212115438900_r60","unstructured":"Siahkoohi\n              A.\n            \n            \n              Orozco\n              R.\n            \n            \n              Rizzuti\n              G.\n            \n            \n              Herrmann\n              F. J.\n            \n          , 2022b, Wave-equation based inversion with amortized variational Bayesian inference: EAGE Deep Learning For Seismic Processing: Investigating the Foundations Workshop."},{"key":"2025121212115438900_r61","doi-asserted-by":"crossref","unstructured":"Siahkoohi\n              A.\n            \n            \n              Rizzuti\n              G.\n            \n            \n              Herrmann\n              F. J.\n            \n          , 2020a, A deep-learning based Bayesian approach to seismic imaging and uncertainty quantification: 82nd Annual International Conference and Exhibition, EAGE, Extended Abstracts, doi: http:\/\/dx.doi.org\/10.3997\/2214-4609.202010770.","DOI":"10.3997\/2214-4609.202010770"},{"key":"2025121212115438900_r62","doi-asserted-by":"crossref","unstructured":"Siahkoohi\n              A.\n            \n            \n              Rizzuti\n              G.\n            \n            \n              Herrmann\n              F. J.\n            \n          , 2020b, Uncertainty quantification in imaging and automatic horizon tracking \u2014 A Bayesian deep-prior based approach: 90th Annual International Meeting, SEG, Expanded Abstracts, 1636\u20131640, doi: http:\/\/dx.doi.org\/10.1190\/segam2020-3417560.1.","DOI":"10.1190\/segam2020-3417560.1"},{"issue":"5","key":"2025121212115438900_r63","doi-asserted-by":"crossref","first-page":"S281","DOI":"10.1190\/geo2021-0666.1","article-title":"Deep Bayesian inference for seismic imaging with tasks","volume":"87","author":"Siahkoohi","year":"2022","journal-title":"Geophysics","ISSN":"https:\/\/id.crossref.org\/issn\/0016-8033","issn-type":"print"},{"key":"2025121212115438900_r64","unstructured":"Siahkoohi\n              A.\n            \n            \n              Rizzuti\n              G.\n            \n            \n              Louboutin\n              M.\n            \n            \n              Witte\n              P.\n            \n            \n              Herrmann\n              F. J.\n            \n          , 2021, Preconditioned training of normalizing flows for variational inference in inverse problems: 3rd Symposium on Advances in Approximate Bayesian Inference."},{"issue":"6","key":"2025121212115438900_r65","doi-asserted-by":"crossref","first-page":"R1003","DOI":"10.1190\/geo2018-0893.1","article-title":"A two-stage Markov chain Monte Carlo method for seismic inversion and uncertainty quantification","volume":"84","author":"Stuart","year":"2019","journal-title":"Geophysics","ISSN":"https:\/\/id.crossref.org\/issn\/0016-8033","issn-type":"print"},{"issue":"3","key":"2025121212115438900_r66","doi-asserted-by":"crossref","first-page":"R303","DOI":"10.1190\/geo2020-0312.1","article-title":"Physics-guided deep learning for seismic inversion with hybrid training and uncertainty analysis","volume":"86","author":"Sun","year":"2021","journal-title":"Geophysics","ISSN":"https:\/\/id.crossref.org\/issn\/0016-8033","issn-type":"print"},{"key":"2025121212115438900_r67","doi-asserted-by":"crossref","DOI":"10.1109\/CDC51059.2022.9992776","article-title":"An optimal transport formulation of Bayes\u2019 law for nonlinear filtering algorithms","author":"Taghvaei","year":"2022"},{"key":"2025121212115438900_r68","doi-asserted-by":"crossref","DOI":"10.1137\/1.9780898717921","volume-title":"Inverse problem theory and methods for model parameter estimation","author":"Tarantola","year":"2005"},{"key":"2025121212115438900_r69","unstructured":"Teshima\n              T.\n            \n            \n              Ishikawa\n              I.\n            \n            \n              Tojo\n              K.\n            \n            \n              Oono\n              K.\n            \n            \n              Ikeda\n              M.\n            \n            \n              Sugiyama\n              M.\n            \n          , 2020, Coupling based invertible neural networks are universal diffeomorphism approximators: Advances in Neural Information Processing Systems, Curran Associates, Inc., 3362\u20133373."},{"key":"2025121212115438900_r70","doi-asserted-by":"crossref","first-page":"840","DOI":"10.1190\/1.1484528","article-title":"Structural uncertainties: Determination, management, and applications","volume":"67","author":"Thore","year":"2002","journal-title":"Geophysics","ISSN":"https:\/\/id.crossref.org\/issn\/0016-8033","issn-type":"print"},{"key":"2025121212115438900_r71","unstructured":"Tieleman\n              T.\n            \n            \n              Hinton\n              G.\n            \n          , 2012, Lecture 6.5-RMSprop: Divide the gradient by a running average of its recent magnitude, https:\/\/www.cs.toronto.edu\/~tijmen\/csc321\/slides\/lecture_slides_lec6.pdf, accessed 2 April 2021."},{"key":"2025121212115438900_r72","unstructured":"T\u00f6lle\n              M.\n            \n            \n              Laves\n              M.-H.\n            \n            \n              Schlaefer\n              A.\n            \n          , 2021, A mean-field variational inference approach to deep image prior for inverse problems in medical imaging: Medical Imaging with Deep Learning."},{"key":"2025121212115438900_r73","unstructured":"Veritas, 2005, Parihaka 3D marine seismic survey \u2014 Acquisition and processing report: Technical Report New Zealand Petroleum Report 3460, New Zealand Petroleum & Minerals, Wellington."},{"key":"2025121212115438900_r74","doi-asserted-by":"crossref","DOI":"10.1007\/978-3-540-71050-9","volume-title":"Optimal transport: Old and new","author":"Villani","year":"2009"},{"key":"2025121212115438900_r75","volume-title":"Graphical models, exponential families, and variational inference","author":"Wainwright","year":"2008"},{"key":"2025121212115438900_r76","unstructured":"Welling\n              M.\n            \n            \n              Teh\n              Y. W.\n            \n          , 2011, Bayesian learning via stochastic gradient Langevin dynamics: Proceedings of the 28th International Conference on Machine Learning, Omnipress, 681\u2013688."},{"key":"2025121212115438900_r77","unstructured":"WesternGeco., 2012, Parihaka 3D PSTM final processing report: Technical Report New Zealand Petroleum Report 4582, New Zealand Petroleum & Minerals, Wellington."},{"key":"2025121212115438900_r78","unstructured":"Whang\n              J.\n            \n            \n              Lindgren\n              E.\n            \n            \n              Dimakis\n              A.\n            \n          , 2021, Composing normalizing flows for inverse problems: International Conference on Machine Learning, 11158\u201311169."},{"key":"2025121212115438900_r79","unstructured":"Witte\n              P.\n            \n            \n              Rizzuti\n              G.\n            \n            \n              Louboutin\n              M.\n            \n            \n              Siahkoohi\n              A.\n            \n            \n              Herrmann\n              F.\n            \n            \n              Peters\n              B.\n            \n          , 2021, InvertibleNetworks.jl: A Julia framework for invertible neural networks, Georgia Institute of Technology, https:\/\/github.com\/slimgroup\/InvertibleNetworks.jl, accessed 6 December 2021."},{"issue":"3","key":"2025121212115438900_r80","doi-asserted-by":"crossref","first-page":"F57","DOI":"10.1190\/geo2018-0174.1","article-title":"A large-scale framework for symbolic implementations of seismic inversion algorithms in Julia","volume":"84","author":"Witte","year":"2019","journal-title":"Geophysics","ISSN":"https:\/\/id.crossref.org\/issn\/0016-8033","issn-type":"print"},{"key":"2025121212115438900_r81","doi-asserted-by":"crossref","unstructured":"Yang\n              Y.\n            \n            \n              Soatto\n              S.\n            \n          , 2018, Conditional prior networks for optical flow: Proceedings of the European Conference on Computer Vision, 271\u2013287.","DOI":"10.1007\/978-3-030-01267-0_17"},{"key":"2025121212115438900_r82","unstructured":"Yosinski\n              J.\n            \n            \n              Clune\n              J.\n            \n            \n              Bengio\n              Y.\n            \n            \n              Lipson\n              H.\n            \n          , 2014, How transferable are features in deep neural networks?Proceedings of the 27th International Conference on Neural Information Processing Systems, 3320\u20133328."},{"key":"2025121212115438900_r83","article-title":"Task agnostic continual learning using online variational Bayes","author":"Zeno","year":"2018"},{"key":"2025121212115438900_r84","article-title":"Seismic tomography using variational inference methods","volume":"125","author":"Zhang","year":"2020","journal-title":"Journal of Geophysical Research: Solid Earth"},{"key":"2025121212115438900_r85","article-title":"Bayesian geophysical inversion using invertible neural networks","volume":"126","author":"Zhang","year":"2021","journal-title":"Journal of Geophysical Research: Solid Earth"},{"key":"2025121212115438900_r86","doi-asserted-by":"crossref","DOI":"10.1093\/gji\/ggad057","article-title":"3D Bayesian variational full waveform inversion","author":"Zhang","year":"2023","journal-title":"Geophysical Journal International"},{"key":"2025121212115438900_r87","first-page":"73","volume-title":"Inversion of geophysical data","author":"Zhang","year":"2021"},{"key":"2025121212115438900_r88","doi-asserted-by":"crossref","first-page":"213","DOI":"10.1093\/gji\/ggab298","article-title":"Bayesian seismic tomography using normalizing flows","volume":"228","author":"Zhao","year":"2021","journal-title":"Geophysical Journal International","ISSN":"https:\/\/id.crossref.org\/issn\/0956-540X","issn-type":"print"},{"key":"2025121212115438900_r89","article-title":"Interrogating subsurface structures using probabilistic tomography: An example assessing the volume of Irish Sea basins","volume":"127","author":"Zhao","year":"2022","journal-title":"Journal of Geophysical Research: Solid Earth"},{"key":"2025121212115438900_r90","doi-asserted-by":"crossref","unstructured":"Zhao\n              Z.\n            \n            \n              Sen\n              M. K.\n            \n          , 2019, A gradient based MCMC method for FWI and uncertainty analysis: 89th Annual International Meeting, SEG, Expanded Abstracts, 1465\u20131469, doi: http:\/\/dx.doi.org\/10.1190\/geo2019-0585.1.","DOI":"10.1190\/segam2019-3216560.1"}],"container-title":["Geophysics"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/pubs.geoscienceworld.org\/seg\/geophysics\/article-pdf\/88\/3\/R297\/5872498\/geo-2022-0472.1.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"syndication"},{"URL":"https:\/\/pubs.geoscienceworld.org\/seg\/geophysics\/article-pdf\/88\/3\/R297\/5872498\/geo-2022-0472.1.pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,12,12]],"date-time":"2025-12-12T17:12:05Z","timestamp":1765559525000},"score":1,"resource":{"primary":{"URL":"https:\/\/pubs.geoscienceworld.org\/geophysics\/article\/88\/3\/R297\/622249\/Reliable-amortized-variational-inference-with"},"secondary":[{"URL":"https:\/\/pubs.geoscienceworld.org\/geophysics\/article\/88\/3\/r297\/622249\/reliable-amortized-variational-inference-with","label":"geoscienceworld"}]},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2023,2,8]]},"references-count":92,"journal-issue":{"issue":"3","published-print":{"date-parts":[[2023,5,1]]}},"URL":"https:\/\/doi.org\/10.1190\/geo2022-0472.1","relation":{},"ISSN":["0016-8033","1942-2156"],"issn-type":[{"value":"0016-8033","type":"print"},{"value":"1942-2156","type":"electronic"}],"subject":[],"published":{"date-parts":[[2023,2,8]]},"assertion":[{"value":"2022-07-23","order":0,"name":"received","label":"Received","group":{"name":"publication_history","label":"Publication History"}},{"value":"2022-12-12","order":1,"name":"revised","label":"Revised","group":{"name":"publication_history","label":"Publication History"}},{"value":"2023-01-14","order":2,"name":"accepted","label":"Accepted","group":{"name":"publication_history","label":"Publication History"}},{"value":"2023-04-13","order":3,"name":"published","label":"Published","group":{"name":"publication_history","label":"Publication History"}}]}}