{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,2,10]],"date-time":"2026-02-10T20:34:07Z","timestamp":1770755647695,"version":"3.50.0"},"reference-count":69,"publisher":"Oxford University Press (OUP)","issue":"1","license":[{"start":{"date-parts":[[2025,7,29]],"date-time":"2025-07-29T00:00:00Z","timestamp":1753747200000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/academic.oup.com\/pages\/standard-publication-reuse-rights"}],"funder":[{"DOI":"10.13039\/100000001","name":"NSF","doi-asserted-by":"publisher","award":["DMS-2210717"],"award-info":[{"award-number":["DMS-2210717"]}],"id":[{"id":"10.13039\/100000001","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/100000001","name":"NSF","doi-asserted-by":"publisher","award":["DMS-2210002"],"award-info":[{"award-number":["DMS-2210002"]}],"id":[{"id":"10.13039\/100000001","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/100000001","name":"NSF","doi-asserted-by":"publisher","award":["DMS-2412833"],"award-info":[{"award-number":["DMS-2412833"]}],"id":[{"id":"10.13039\/100000001","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"published-print":{"date-parts":[[2026,2,10]]},"abstract":"<jats:title>Abstract<\/jats:title>\n                  <jats:p>This article addresses the problem of testing the conditional independence of two generic random vectors X and Y given a third random vector Z, which plays an important role in statistical and machine learning applications. We propose a new non-parametric testing procedure that avoids explicitly estimating any conditional distributions but instead requires sampling from the two marginal conditional distributions of X given Z and Y given Z. We further propose using a generative neural network (GNN) framework to sample from these approximated marginal conditional distributions, which tends to mitigate the curse of dimensionality due to its adaptivity to any low-dimensional structures and smoothness underlying the data. Theoretically, our test statistic is shown to enjoy a double robustness property against GNN approximation errors, meaning that the test statistic retains all desirable properties of the oracle test statistic utilizing the true marginal conditional distributions, as long as the product of the two approximation errors decays to zero faster than the parametric rate. Asymptotic properties of our statistic and the consistency of a bootstrap procedure are derived under both null and local alternatives. Extensive numerical experiments and real data analysis illustrate the effectiveness and broad applicability of our proposed test.<\/jats:p>","DOI":"10.1093\/jrsssb\/qkaf047","type":"journal-article","created":{"date-parts":[[2025,7,4]],"date-time":"2025-07-04T07:56:49Z","timestamp":1751615809000},"page":"195-220","source":"Crossref","is-referenced-by-count":0,"title":["Doubly robust conditional independence testing with generative neural networks"],"prefix":"10.1093","volume":"88","author":[{"given":"Yi","family":"Zhang","sequence":"first","affiliation":[{"name":"Department of Statistics, University of Illinois at Urbana-Champaign , Champaign ,","place":["USA"]}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Linjun","family":"Huang","sequence":"additional","affiliation":[{"name":"Department of Statistics, University of Illinois at Urbana-Champaign , Champaign ,","place":["USA"]}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Yun","family":"Yang","sequence":"additional","affiliation":[{"name":"Department of Mathematics, University of Maryland at College Park , College Park ,","place":["USA"]}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Xiaofeng","family":"Shao","sequence":"additional","affiliation":[{"name":"Department of Statistics and Data Science, and Department of Economics, Washington University in St Louis , St. Louis ,","place":["USA"]}],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"286","published-online":{"date-parts":[[2025,7,29]]},"reference":[{"issue":"6","key":"2026021010563390900_qkaf047-B1","doi-asserted-by":"publisher","first-page":"3070","DOI":"10.1214\/21-AOS2073","article-title":"A simple measure of conditional dependence","volume":"49","author":"Azadkia","year":"2021","journal-title":"Annals of Statistics"},{"key":"2026021010563390900_qkaf047-B2","doi-asserted-by":"publisher","author":"Banerjee","DOI":"10.48550\/arXiv.2407.21456"},{"issue":"4","key":"2026021010563390900_qkaf047-B3","doi-asserted-by":"publisher","first-page":"962","DOI":"10.1111\/biom.2005.61.issue-4","article-title":"Doubly robust estimation in missing data and causal inference models","volume":"61","author":"Bang","year":"2005","journal-title":"Biometrics"},{"issue":"7391","key":"2026021010563390900_qkaf047-B4","doi-asserted-by":"publisher","first-page":"603","DOI":"10.1038\/nature11003","article-title":"The cancer cell line encyclopedia enables predictive modelling of anticancer drug sensitivity","volume":"483","author":"Barretina","year":"2012","journal-title":"Nature"},{"issue":"3","key":"2026021010563390900_qkaf047-B5","doi-asserted-by":"publisher","first-page":"279","DOI":"10.1016\/S0167-9473(00)00046-3","article-title":"Bandwidth selection for kernel conditional density estimation","volume":"36","author":"Bashtannyk","year":"2001","journal-title":"Computational Statistics & Data Analysis"},{"key":"2026021010563390900_qkaf047-B6","doi-asserted-by":"publisher","first-page":"13998","DOI":"10.48550\/arXiv.1910.06221","article-title":"Conditional independence testing using generative adversarial networks","volume":"32","author":"Bellot","year":"2019","journal-title":"Advances in Neural Information Processing Systems"},{"issue":"2","key":"2026021010563390900_qkaf047-B7","doi-asserted-by":"publisher","first-page":"28","DOI":"10.1007\/s10462-023-10662-6","article-title":"Autoencoders and their applications in machine learning: A survey","volume":"57","author":"Berahmand","year":"2024","journal-title":"Artificial Intelligence Review"},{"issue":"20","key":"2026021010563390900_qkaf047-B8","doi-asserted-by":"publisher","first-page":"2470","DOI":"10.3390\/electronics10202470","article-title":"Cnn variants for computer vision: History, architecture, application, challenges and future scope","volume":"10","author":"Bhatt","year":"2021","journal-title":"Electronics"},{"key":"2026021010563390900_qkaf047-B9","doi-asserted-by":"publisher","author":"Bonheme","DOI":"10.48550\/arXiv.2209.12806"},{"issue":"85","key":"2026021010563390900_qkaf047-B10","doi-asserted-by":"publisher","first-page":"1","DOI":"10.48550\/arXiv.2006.04176","article-title":"A distribution free conditional independence test with applications to causal discovery","volume":"23","author":"Cai","year":"2022","journal-title":"Journal of Machine Learning Research"},{"issue":"3","key":"2026021010563390900_qkaf047-B11","doi-asserted-by":"publisher","first-page":"551","DOI":"10.1111\/rssb.12265","article-title":"Panning for gold: \u2018model-x\u2019 knockoffs for high dimensional controlled variable selection","volume":"80","author":"Candes","year":"2018","journal-title":"Journal of the Royal Statistical Society: Series B, Statistical Methodology"},{"key":"2026021010563390900_qkaf047-B12","doi-asserted-by":"publisher","author":"Chen","DOI":"10.48550\/arXiv.2002.03938"},{"issue":"2","key":"2026021010563390900_qkaf047-B13","doi-asserted-by":"publisher","first-page":"455","DOI":"10.1214\/aos\/1021379861","article-title":"Dimension reduction for conditional mean in regression","volume":"30","author":"Cook","year":"2002","journal-title":"Annals of Statistics"},{"issue":"3","key":"2026021010563390900_qkaf047-B14","doi-asserted-by":"publisher","first-page":"581","DOI":"10.1093\/biomet\/67.3.581","article-title":"Partial association measures and an application to qualitative regression","volume":"67","author":"Daudin","year":"1980","journal-title":"Biometrika"},{"key":"2026021010563390900_qkaf047-B15","author":"Dom\u00ednguez"},{"key":"2026021010563390900_qkaf047-B16","doi-asserted-by":"publisher","author":"Dziugaite","DOI":"10.48550\/arXiv.1505.03906"},{"issue":"1","key":"2026021010563390900_qkaf047-B17","doi-asserted-by":"publisher","first-page":"100645","DOI":"10.1016\/j.cosrev.2024.100645","article-title":"Backbones-review: Feature extractor networks for deep learning and deep reinforcement learning approaches in computer vision","volume":"53","author":"Elharrouss","year":"2024","journal-title":"Computer Science Review"},{"key":"2026021010563390900_qkaf047-B18","author":"Fournier"},{"key":"2026021010563390900_qkaf047-B19","first-page":"873","article-title":"Kernel measures of conditional dependence","volume":"20","author":"Fukumizu","year":"2007","journal-title":"Advances in Neural Information Processing Systems"},{"issue":"7","key":"2026021010563390900_qkaf047-B20","doi-asserted-by":"publisher","first-page":"761","DOI":"10.1093\/aje\/kwq439","article-title":"Doubly robust estimation of causal effects","volume":"173","author":"Funk","year":"2011","journal-title":"American Journal of Epidemiology"},{"key":"2026021010563390900_qkaf047-B21","author":"Geiger"},{"issue":"452","key":"2026021010563390900_qkaf047-B22","doi-asserted-by":"publisher","first-page":"1304","DOI":"10.1080\/01621459.2000.10474336","article-title":"The variable selection problem","volume":"95","author":"George","year":"2000","journal-title":"Journal of the American Statistical Association"},{"key":"2026021010563390900_qkaf047-B23","doi-asserted-by":"publisher","first-page":"29343","DOI":"10.48550\/arXiv.2308.10349","article-title":"Battle of the backbones: A large-scale comparison of pretrained models across computer vision tasks","volume":"36","author":"Goldblum","year":"2023","journal-title":"Advances in Neural Information Processing Systems"},{"key":"2026021010563390900_qkaf047-B24","doi-asserted-by":"publisher","first-page":"2672","DOI":"10.48550\/arXiv.1406.2661","article-title":"Generative adversarial nets","volume":"27","author":"Goodfellow","year":"2014","journal-title":"Advances in Neural Information Processing Systems"},{"issue":"25","key":"2026021010563390900_qkaf047-B25","doi-asserted-by":"publisher","first-page":"723","DOI":"10.5555\/2188385.2188410","article-title":"A kernel two-sample test","volume":"13","author":"Gretton","year":"2012","journal-title":"Journal of Machine Learning Research"},{"key":"2026021010563390900_qkaf047-B26","author":"Gretton"},{"key":"2026021010563390900_qkaf047-B27","author":"He"},{"key":"2026021010563390900_qkaf047-B28","author":"Hore"},{"issue":"5","key":"2026021010563390900_qkaf047-B29","doi-asserted-by":"publisher","first-page":"3583","DOI":"10.1109\/TIT.2023.3340658","article-title":"Deep dimension reduction for supervised representation learning","volume":"70","author":"Huang","year":"2024","journal-title":"IEEE Transactions on Information Theory"},{"issue":"216","key":"2026021010563390900_qkaf047-B30","doi-asserted-by":"publisher","first-page":"1","DOI":"10.48550\/arXiv.2103.03681","article-title":"Kernel partial correlation coefficient\u2014A measure of conditional dependence","volume":"23","author":"Huang","year":"2022","journal-title":"Journal of Machine Learning Research"},{"issue":"7","key":"2026021010563390900_qkaf047-B31","doi-asserted-by":"publisher","first-page":"5929","DOI":"10.1007\/s10462-022-10305-2","article-title":"Image denoising in the deep learning era","volume":"56","author":"Izadi","year":"2023","journal-title":"Artificial Intelligence Review"},{"issue":"4","key":"2026021010563390900_qkaf047-B32","doi-asserted-by":"publisher","first-page":"1297","DOI":"10.1080\/10618600.2015.1094393","article-title":"Nonparametric conditional density estimation in a high-dimensional regression setting","volume":"25","author":"Izbicki","year":"2016","journal-title":"Journal of Computational and Graphical Statistics"},{"key":"2026021010563390900_qkaf047-B33","doi-asserted-by":"publisher","author":"Jiang","DOI":"10.48550\/arXiv.1806.01054"},{"key":"2026021010563390900_qkaf047-B34","doi-asserted-by":"crossref","DOI":"10.1007\/978-1-4757-4015-8","volume-title":"Foundations of modern probability","author":"Kallenberg","year":"2002","edition":"2nd ed."},{"issue":"7","key":"2026021010563390900_qkaf047-B35","doi-asserted-by":"publisher","first-page":"1462","DOI":"10.1162\/neco_a_01195","article-title":"Effect of depth and width on local minima in deep learning","volume":"31","author":"Kawaguchi","year":"2019","journal-title":"Neural Computation"},{"issue":"6","key":"2026021010563390900_qkaf047-B36","doi-asserted-by":"publisher","first-page":"3388","DOI":"10.1214\/22-AOS2233","article-title":"Local permutation tests for conditional independence","volume":"50","author":"Kim","year":"2022","journal-title":"Annals of Statistics"},{"key":"2026021010563390900_qkaf047-B37","volume-title":"Probabilistic graphical models: Principles and techniques","author":"Koller","year":"2009"},{"key":"2026021010563390900_qkaf047-B38","doi-asserted-by":"publisher","first-page":"396","DOI":"10.5555\/109230.109298","article-title":"Handwritten digit recognition with a back-propagation network","volume":"2","author":"LeCun","year":"1989","journal-title":"Advances in Neural Information Processing Systems"},{"key":"2026021010563390900_qkaf047-B39","doi-asserted-by":"crossref","DOI":"10.1201\/9781315119427","volume-title":"Sufficient dimension reduction: Methods and applications with R","author":"Li","year":"2018"},{"issue":"3","key":"2026021010563390900_qkaf047-B40","doi-asserted-by":"publisher","first-page":"e1489","DOI":"10.1002\/wics.v12.3","article-title":"On nonparametric conditional independence tests for continuous variables","volume":"12","author":"Li","year":"2020","journal-title":"Wiley Interdisciplinary Reviews: Computational Statistics"},{"issue":"2","key":"2026021010563390900_qkaf047-B41","doi-asserted-by":"publisher","first-page":"3937","DOI":"10.1214\/22-EJS2037","article-title":"Minimax optimal conditional density estimation under total variation smoothness","volume":"16","author":"Li","year":"2022","journal-title":"Electronic Journal of Statistics"},{"key":"2026021010563390900_qkaf047-B42","author":"Li"},{"key":"2026021010563390900_qkaf047-B43","author":"Li"},{"key":"2026021010563390900_qkaf047-B44","author":"Long"},{"issue":"1-2","key":"2026021010563390900_qkaf047-B45","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1561\/2200000060","article-title":"Kernel mean embedding of distributions: A review and beyond","volume":"10","author":"Muandet","year":"2017","journal-title":"Foundations and Trends\u00ae in Machine Learning"},{"issue":"4","key":"2026021010563390900_qkaf047-B46","doi-asserted-by":"publisher","first-page":"2151","DOI":"10.1214\/20-AOS2030","article-title":"Minimax optimal conditional independence testing","volume":"49","author":"Neykov","year":"2021","journal-title":"Annals of Statistics"},{"issue":"3","key":"2026021010563390900_qkaf047-B47","doi-asserted-by":"publisher","first-page":"895","DOI":"10.1214\/24-AOS2372","article-title":"Reconciling model-x and doubly robust approaches to conditional independence testing","volume":"52","author":"Niu","year":"2024","journal-title":"Annals of Statistics"},{"issue":"3","key":"2026021010563390900_qkaf047-B48","doi-asserted-by":"publisher","first-page":"1109","DOI":"10.1214\/17-AOS1579","article-title":"Ball divergence: Nonparametric two sample test","volume":"46","author":"Pan","year":"2018","journal-title":"Annals of Statistics"},{"key":"2026021010563390900_qkaf047-B49","doi-asserted-by":"publisher","first-page":"17104","DOI":"10.48550\/arXiv.2006.05433","article-title":"A measure-theoretic approach to kernel conditional mean embeddings","volume":"33","author":"Park","year":"2020","journal-title":"Advances in Neural Information Processing Systems"},{"key":"2026021010563390900_qkaf047-B50","doi-asserted-by":"crossref","DOI":"10.1017\/CBO9780511803161","volume-title":"Causality: Models, reasoning and inference","author":"Pearl","year":"2009"},{"key":"2026021010563390900_qkaf047-B51","doi-asserted-by":"publisher","author":"Pogodin","DOI":"10.48550\/arXiv.2212.08645"},{"key":"2026021010563390900_qkaf047-B52","doi-asserted-by":"publisher","author":"Pogodin","DOI":"10.48550\/arXiv.2402.13196"},{"key":"2026021010563390900_qkaf047-B53","author":"Ronneberger"},{"key":"2026021010563390900_qkaf047-B54","author":"Scetbon"},{"key":"2026021010563390900_qkaf047-B55","doi-asserted-by":"publisher","first-page":"3039","DOI":"10.48550\/arXiv.1709.06180","article-title":"Model-powered conditional independence test","volume":"30","author":"Sen","year":"2017","journal-title":"Advances in Neural Information Processing Systems"},{"issue":"3","key":"2026021010563390900_qkaf047-B56","doi-asserted-by":"publisher","first-page":"1514","DOI":"10.1214\/19-AOS1857","article-title":"The hardness of conditional independence testing and the generalised covariance measure","volume":"48","author":"Shah","year":"2020","journal-title":"Annals of Statistics"},{"issue":"7","key":"2026021010563390900_qkaf047-B57","doi-asserted-by":"publisher","first-page":"1","DOI":"10.48550\/arXiv.1912.01103","article-title":"On distance and kernel measures of conditional dependence","volume":"24","author":"Sheng","year":"2023","journal-title":"Journal of Machine Learning Research"},{"issue":"285","key":"2026021010563390900_qkaf047-B58","doi-asserted-by":"publisher","first-page":"1","DOI":"10.48550\/arXiv.2106.07229","article-title":"Double generative adversarial networks for conditional independence testing","volume":"22","author":"Shi","year":"2021","journal-title":"Journal of Machine Learning Research"},{"issue":"70","key":"2026021010563390900_qkaf047-B59","doi-asserted-by":"publisher","first-page":"2389","DOI":"10.48550\/arXiv.1003.0887","article-title":"Universality, characteristic kernels and rkhs embedding of measures","volume":"12","author":"Sriperumbudur","year":"2011","journal-title":"Journal of Machine Learning Research"},{"issue":"2","key":"2026021010563390900_qkaf047-B60","doi-asserted-by":"publisher","first-page":"807","DOI":"10.1016\/j.jeconom.2006.11.006","article-title":"A consistent characteristic function-based test for conditional independence","volume":"141","author":"Su","year":"2007","journal-title":"Journal of Econometrics"},{"issue":"1","key":"2026021010563390900_qkaf047-B61","doi-asserted-by":"publisher","first-page":"27","DOI":"10.1016\/j.jeconom.2014.04.006","article-title":"Testing conditional independence via empirical likelihood","volume":"182","author":"Su","year":"2014","journal-title":"Journal of Econometrics"},{"issue":"233","key":"2026021010563390900_qkaf047-B62","doi-asserted-by":"publisher","first-page":"1","DOI":"10.48550\/arXiv.1703.01950","article-title":"Characteristic and universal tensor product kernels","volume":"18","author":"Szab\u00f3","year":"2018","journal-title":"Journal of Machine Learning Research"},{"issue":"3","key":"2026021010563390900_qkaf047-B63","doi-asserted-by":"publisher","first-page":"1282","DOI":"10.1214\/23-AOS2291","article-title":"Minimax rate of distribution estimation on unknown submanifolds under adversarial losses","volume":"51","author":"Tang","year":"2023","journal-title":"Annals of Statistics"},{"issue":"512","key":"2026021010563390900_qkaf047-B64","doi-asserted-by":"publisher","first-page":"1726","DOI":"10.1080\/01621459.2014.993081","article-title":"Conditional distance correlation","volume":"110","author":"Wang","year":"2015","journal-title":"Journal of the American Statistical Association"},{"issue":"17","key":"2026021010563390900_qkaf047-B65","doi-asserted-by":"publisher","first-page":"8643","DOI":"10.3390\/app12178643","article-title":"A comparison of pooling methods for convolutional neural networks","volume":"12","author":"Zafar","year":"2022","journal-title":"Applied Sciences"},{"key":"2026021010563390900_qkaf047-B66","author":"Zhang"},{"issue":"1","key":"2026021010563390900_qkaf047-B67","doi-asserted-by":"publisher","first-page":"219","DOI":"10.1214\/17-AOS1548","article-title":"Conditional mean and quantile dependence testing in high dimension","volume":"46","author":"Zhang","year":"2018","journal-title":"Annals of Statistics"},{"issue":"543","key":"2026021010563390900_qkaf047-B68","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1080\/01621459.2021.1906684","article-title":"A deep generative approach to conditional sampling","volume":"118","author":"Zhou","year":"2022","journal-title":"Journal of the American Statistical Association"},{"issue":"4","key":"2026021010563390900_qkaf047-B69","doi-asserted-by":"publisher","first-page":"1204","DOI":"10.1093\/jrsssb\/qkad064","article-title":"Testing for the Markov property in time series via deep conditional generative learning","volume":"85","author":"Zhou","year":"2023","journal-title":"Journal of the Royal Statistical Society: Series B, Statistical Methodology"}],"container-title":["Journal of the Royal Statistical Society Series B: Statistical Methodology"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/academic.oup.com\/jrsssb\/article-pdf\/88\/1\/195\/63879792\/qkaf047.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"syndication"},{"URL":"https:\/\/academic.oup.com\/jrsssb\/article-pdf\/88\/1\/195\/63879792\/qkaf047.pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,2,10]],"date-time":"2026-02-10T15:56:42Z","timestamp":1770739002000},"score":1,"resource":{"primary":{"URL":"https:\/\/academic.oup.com\/jrsssb\/article\/88\/1\/195\/8217135"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,7,29]]},"references-count":69,"journal-issue":{"issue":"1","published-online":{"date-parts":[[2025,7,29]]},"published-print":{"date-parts":[[2026,2,10]]}},"URL":"https:\/\/doi.org\/10.1093\/jrsssb\/qkaf047","relation":{},"ISSN":["1369-7412","1467-9868"],"issn-type":[{"value":"1369-7412","type":"print"},{"value":"1467-9868","type":"electronic"}],"subject":[],"published-other":{"date-parts":[[2026,2]]},"published":{"date-parts":[[2025,7,29]]}}}