{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,11,2]],"date-time":"2026-11-02T23:04:14Z","timestamp":1793660654186,"version":"4.5.1"},"reference-count":282,"publisher":"Institute of Electrical and Electronics Engineers (IEEE)","issue":"5","license":[{"start":{"date-parts":[[2021,5,1]],"date-time":"2021-05-01T00:00:00Z","timestamp":1619827200000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/legalcode"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["Proc. IEEE"],"published-print":{"date-parts":[[2021,5]]},"DOI":"10.1109\/jproc.2021.3058954","type":"journal-article","created":{"date-parts":[[2021,2,26]],"date-time":"2021-02-26T15:39:32Z","timestamp":1614353972000},"page":"612-634","source":"Crossref","is-referenced-by-count":1112,"title":["Toward Causal Representation Learning"],"prefix":"10.1109","volume":"109","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-8177-0925","authenticated-orcid":false,"given":"Bernhard","family":"Scholkopf","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-4850-0683","authenticated-orcid":false,"given":"Francesco","family":"Locatello","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-1712-060X","authenticated-orcid":false,"given":"Stefan","family":"Bauer","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Nan Rosemary","family":"Ke","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Nal","family":"Kalchbrenner","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Anirudh","family":"Goyal","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-9322-3515","authenticated-orcid":false,"given":"Yoshua","family":"Bengio","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"263","reference":[{"key":"ref275","first-page":"2037","article-title":"Fairness in decision-making—The causal explanation formula","author":"zhang","year":"2018","journal-title":"Proc 32nd AAAI Conf Artif Intell"},{"key":"ref274","article-title":"Deep reinforcement learning with relational inductive biases","author":"zambaldi","year":"2018","journal-title":"Proc Int Conf Learn Represent"},{"key":"ref277","first-page":"3150","article-title":"Multi-source domain adaptation: A causal view","author":"zhang","year":"2015","journal-title":"Proc 29th AAAI Conf Artif Intell"},{"key":"ref276","first-page":"13401","article-title":"Near-optimal reinforcement learning in dynamic treatment regimes","author":"zhang","year":"2019","journal-title":"Proc Adv Neural Inf Process Syst"},{"key":"ref271","doi-asserted-by":"publisher","DOI":"10.1109\/ICRA.2016.7487172"},{"key":"ref270","doi-asserted-by":"publisher","DOI":"10.1162\/089976602317318938"},{"key":"ref170","article-title":"Benchmarking robustness in object detection: Autonomous driving when winter is coming","author":"michaelis","year":"2019","journal-title":"arXiv 1907 07484"},{"key":"ref273","article-title":"GANITE: Estimation of individualized treatment effects using generative adversarial nets","author":"yoon","year":"2018","journal-title":"Proc Int Conf Learn Represent"},{"key":"ref272","article-title":"CLEVRER: CoLlision events for video REpresentation and reasoning","author":"yi","year":"2019","journal-title":"arXiv 1910 01442"},{"key":"ref172","doi-asserted-by":"publisher","DOI":"10.1088\/0004-637X\/809\/1\/25"},{"key":"ref171","doi-asserted-by":"publisher","DOI":"10.1038\/nature14236"},{"key":"ref174","first-page":"639","article-title":"On causal discovery with cyclic additive noise models","author":"mooij","year":"2011","journal-title":"Proc Adv Neural Inf Process Syst (NIPS)"},{"key":"ref173","first-page":"440","article-title":"From ordinary differential equations to structural causal models: The deterministic case","author":"mooij","year":"2013","journal-title":"Proc 29th Conf Annu Conf Uncertainty Artif Intell"},{"key":"ref176","first-page":"1","article-title":"Distinguishing cause from effect using observational data: Methods and benchmarks","volume":"17","author":"mooij","year":"2016","journal-title":"J Mach Learn Res"},{"key":"ref175","first-page":"745","article-title":"Regression by dependence minimization and its application to causal inference","author":"mooij","year":"2009","journal-title":"Proc 26th Int Conf Mach Learn (ICML)"},{"key":"ref178","first-page":"2863","article-title":"Action-conditional video prediction using deep networks in Atari games","author":"oh","year":"2015","journal-title":"Proc Adv Neural Inf Process Syst"},{"key":"ref177","first-page":"8799","article-title":"Flexible neural representation for physics prediction","author":"mrowca","year":"2018","journal-title":"Proc Adv Neural Inf Process Syst"},{"key":"ref168","doi-asserted-by":"publisher","DOI":"10.1111\/1467-9639.00013"},{"key":"ref169","doi-asserted-by":"publisher","DOI":"10.1109\/DSW.2018.8439889"},{"key":"ref39","first-page":"375","article-title":"Improving the accuracy and speed of support vector learning machines","volume":"9","author":"burges","year":"1997","journal-title":"Advances in neural information processing systems"},{"key":"ref38","article-title":"Woulda, coulda, shoulda: Counterfactually-guided policy search","author":"buesing","year":"2018","journal-title":"arXiv 1811 06272"},{"key":"ref33","doi-asserted-by":"publisher","DOI":"10.1145\/279943.279962"},{"key":"ref32","article-title":"Error asymmetry in causal and anticausal regression","author":"bl\u00f6baum","year":"2016","journal-title":"arXiv 1610 03263"},{"key":"ref31","article-title":"Time series deconfounder: Estimating treatment effects over time in the presence of hidden confounders","author":"bica","year":"2019","journal-title":"arXiv 1902 00450"},{"key":"ref30","article-title":"Counterfactuals uncover the modular structure of deep generative models","author":"besserve","year":"2018","journal-title":"arXiv 1812 03253"},{"key":"ref267","doi-asserted-by":"publisher","DOI":"10.1109\/PRNI.2014.6858551"},{"key":"ref37","doi-asserted-by":"publisher","DOI":"10.1109\/ICDM.2016.0015"},{"key":"ref268","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-642-27645-3_19"},{"key":"ref36","article-title":"Language models are few-shot learners","author":"brown","year":"2020","journal-title":"arXiv 2005 14165"},{"key":"ref269","doi-asserted-by":"publisher","DOI":"10.1111\/cgf.13620"},{"key":"ref35","first-page":"3207","article-title":"Counterfactual reasoning and learning systems: The example of computational advertising","volume":"14","author":"bottou","year":"2013","journal-title":"J Mach Learn Res"},{"key":"ref34","article-title":"Learning first-order symbolic representations for planning from the structure of the state space","author":"bonet","year":"2019","journal-title":"arXiv 1909 05546"},{"key":"ref181","article-title":"Learning explanations that are hard to vary","author":"parascandolo","year":"2021","journal-title":"Proc Int Conf Learn Represent"},{"key":"ref180","first-page":"4036","article-title":"Learning independent causal mechanisms","volume":"80","author":"parascandolo","year":"2018","journal-title":"Proc 35th Int Conf Mach Learn (PMLR)"},{"key":"ref282","article-title":"Making convolutional networks shift-invariant again","author":"zhang","year":"2019","journal-title":"arXiv 1904 11486"},{"key":"ref281","first-page":"819","article-title":"Domain adaptation under target and conditional shift","author":"zhang","year":"2013","journal-title":"Proc 30th Int Conf Mach Learn (ICML)"},{"key":"ref185","article-title":"External validity: From do-calculus to transportability across populations","author":"pearl","year":"2015","journal-title":"arXiv 1503 01603"},{"key":"ref280","first-page":"804","article-title":"Kernel-based conditional independence test and application in causal discovery","author":"zhang","year":"2011","journal-title":"Proc 27th Annu Conf Uncertainty Artif Intell (UAI)"},{"key":"ref184","article-title":"Giving computers free will","author":"pearl","year":"2009"},{"key":"ref183","doi-asserted-by":"publisher","DOI":"10.1017\/CBO9780511803161"},{"key":"ref182","article-title":"Learning independent causal mechanisms","author":"parascandolo","year":"2017","journal-title":"Proc Workshop Learn Disentangled Represent From Perception Control 31st Conf Neural Inf Process Syst (NIPS)"},{"key":"ref189","first-page":"589","article-title":"Identifiability of causal graphs using functional models","author":"peters","year":"2011","journal-title":"Proc 27th Annu Conf Uncertainty Artif Intell (UAI)"},{"key":"ref188","author":"peters","year":"2017","journal-title":"Elements of Causal Inference—Foundations and Learning Algorithms"},{"key":"ref187","doi-asserted-by":"publisher","DOI":"10.1111\/rssb.12167"},{"key":"ref186","article-title":"Causal models for dynamical systems","author":"peters","year":"2020","journal-title":"arXiv 2001 06208"},{"key":"ref28","first-page":"557","article-title":"Group invariance principles for causal generative models","author":"besserve","year":"2018","journal-title":"Proc 21st Int Conf Artif Intell Statist (AISTATS)"},{"key":"ref27","article-title":"Dota 2 with large scale deep reinforcement learning","year":"2019","journal-title":"arXiv 1912 06680"},{"key":"ref179","article-title":"Representation learning with contrastive predictive coding","author":"van den oord","year":"2018","journal-title":"arXiv 1807 03748"},{"key":"ref29","article-title":"A theory of independent mechanisms for extrapolation in generative models","author":"besserve","year":"2021","journal-title":"Proc 35th AAAI Conf Artif Intell Virtual Conf"},{"key":"ref20","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-030-01270-0_28"},{"key":"ref22","article-title":"Independently controllable features","author":"bengio","year":"2017","journal-title":"arXiv 1703 07718"},{"key":"ref21","first-page":"129","article-title":"Impossibility theorems for domain adaptation","author":"ben-david","year":"2010","journal-title":"Proc Int Conf Artif Intell Statist (AISTATS)"},{"key":"ref24","article-title":"Representation learning: A review and new perspectives","author":"bengio","year":"2012","journal-title":"arXiv 1206 5538"},{"key":"ref23","doi-asserted-by":"publisher","DOI":"10.1109\/IJCNN.1991.155621"},{"key":"ref278","doi-asserted-by":"publisher","DOI":"10.24963\/ijcai.2017\/187"},{"key":"ref26","article-title":"Water vapor on the habitable-zone exoplanet K2-18b","author":"benneke","year":"2019","journal-title":"arXiv 1909 04642"},{"key":"ref279","first-page":"647","article-title":"On the identifiability of the post-nonlinear causal model","author":"zhang","year":"2009","journal-title":"Proc 25th Annu Conf Uncertainty Artif Intell (UAI)"},{"key":"ref25","article-title":"A meta-transfer objective for learning to disentangle causal mechanisms","author":"bengio","year":"2019","journal-title":"arXiv 1901 10912"},{"key":"ref50","first-page":"143","article-title":"Inferring deterministic causal relations","author":"daniu\u0161is","year":"2010","journal-title":"Proc 26th Annu Conf Uncertainty Artif Intell (UAI)"},{"key":"ref51","article-title":"Causal reasoning from meta-reinforcement learning","author":"dasgupta","year":"2019","journal-title":"arXiv 1901 08162"},{"key":"ref154","first-page":"6665","article-title":"Fast autoaugment","author":"lim","year":"2019","journal-title":"Proc Adv Neural Inf Process Syst"},{"key":"ref153","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-030-01267-0_38"},{"key":"ref156","article-title":"Detecting and correcting for label shift with black box predictors","author":"lipton","year":"2018","journal-title":"arXiv 1802 03916"},{"key":"ref155","article-title":"Space: Unsupervised object-oriented scene representation via spatial attention and decomposition","author":"lin","year":"2019","journal-title":"Proc Int Conf Learn Represent"},{"key":"ref150","article-title":"Offline reinforcement learning: Tutorial, review, and perspectives on open problems","author":"levine","year":"2020","journal-title":"arXiv 2005 01643"},{"key":"ref152","article-title":"Domain generalization via conditional invariant representation","author":"li","year":"2018","journal-title":"arXiv 1807 08479"},{"key":"ref151","doi-asserted-by":"publisher","DOI":"10.2307\/2025310"},{"key":"ref146","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-642-27645-3_2"},{"key":"ref147","doi-asserted-by":"crossref","DOI":"10.1093\/oso\/9780198522195.001.0001","author":"lauritzen","year":"1996","journal-title":"Graphical Models"},{"key":"ref148","doi-asserted-by":"crossref","first-page":"436","DOI":"10.1038\/nature14539","article-title":"Deep learning","volume":"521","author":"lecun","year":"2015","journal-title":"Nature"},{"key":"ref149","article-title":"Structural autoencoders improve representations for generation and transfer","author":"leeb","year":"2020","journal-title":"arXiv 2006 07796"},{"key":"ref59","doi-asserted-by":"publisher","DOI":"10.1145\/1390156.1390187"},{"key":"ref58","article-title":"On the transfer of disentangled representations in realistic settings","author":"dittadi","year":"2021","journal-title":"Proc Int Conf Learn Represent"},{"key":"ref57","article-title":"Jukebox: A generative model for music","author":"dhariwal","year":"2020","journal-title":"arXiv 2005 00341"},{"key":"ref56","article-title":"A probabilistic theory of pattern recognition","volume":"31","author":"devroye","year":"1996","journal-title":"Applications of Mathematics"},{"key":"ref55","article-title":"BERT: Pre-training of deep bidirectional transformers for language understanding","author":"devlin","year":"2018","journal-title":"arXiv 1810 04805"},{"key":"ref54","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2009.5206848"},{"key":"ref53","author":"dehaene","year":"2020","journal-title":"How We Learn Why Brains Learn Better Than Any Machine … for Now"},{"key":"ref52","doi-asserted-by":"publisher","DOI":"10.1111\/j.2517-6161.1979.tb01052.x"},{"key":"ref40","article-title":"MONet: Unsupervised scene decomposition and representation","author":"burgess","year":"2019","journal-title":"arXiv 1901 11390"},{"key":"ref167","first-page":"10869","article-title":"Domain adaptation by using causal inference to predict invariant conditional distributions","author":"magliacane","year":"2018","journal-title":"Proc NeurIPS"},{"key":"ref166","doi-asserted-by":"publisher","DOI":"10.1016\/j.zemedi.2018.11.002"},{"key":"ref165","article-title":"Deconfounding reinforcement learning in observational settings","author":"lu","year":"2018","journal-title":"arXiv 1812 10576"},{"key":"ref164","article-title":"Sample-efficient reinforcement learning via counterfactual-based data augmentation","author":"lu","year":"2020","journal-title":"arXiv 2012 09092"},{"key":"ref163","author":"lorenz","year":"1973","journal-title":"Die Rückseite des Spiegels"},{"key":"ref162","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2017.14"},{"key":"ref161","first-page":"1452","article-title":"Towards a learning theory of cause-effect inference","author":"lopez-paz","year":"2015","journal-title":"Proc 32nd Int Conf Mach Learn"},{"key":"ref160","article-title":"Object-centric learning with slot attention","author":"locatello","year":"2020","journal-title":"Proc Adv Neural Inf Process Syst"},{"key":"ref4","doi-asserted-by":"publisher","DOI":"10.1093\/oxfordjournals.oep.a041889"},{"key":"ref3","first-page":"129","article-title":"Limits of estimating heterogeneous treatment effects: Guidelines for practical algorithm design","author":"alaa","year":"2018","journal-title":"Proc Int Conf Mach Learn"},{"key":"ref6","article-title":"Deep-treat: Learning optimal personalized treatments from observational data using neural networks","author":"atan","year":"2018","journal-title":"Proc 32nd AAAI Conf Artif Intell"},{"key":"ref5","article-title":"Invariant risk minimization","author":"arjovsky","year":"2019","journal-title":"arXiv 1907 02893"},{"key":"ref159","first-page":"6348","article-title":"Weakly-supervised disentanglement without compromises","author":"locatello","year":"2020","journal-title":"Proc 37th Int Conf Mach Learn (ICML)"},{"key":"ref8","doi-asserted-by":"publisher","DOI":"10.1109\/ISIT44484.2020.9174424"},{"key":"ref7","first-page":"1","article-title":"Why do deep convolutional networks generalize so poorly to small image transformations?","volume":"20","author":"azulay","year":"2019","journal-title":"J Mach Learn Res"},{"key":"ref49","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2019.00020"},{"key":"ref157","first-page":"14544","article-title":"On the fairness of disentangled representations","author":"locatello","year":"2019","journal-title":"Proc Adv Neural Inf Process Syst"},{"key":"ref158","first-page":"4114","article-title":"Challenging common assumptions in the unsupervised learning of disentangled representations","author":"locatello","year":"2019","journal-title":"Proc 36th Int Conf Mach Learn"},{"key":"ref9","article-title":"Systematic generalization: What is required and can it be learned","author":"bahdanau","year":"2018","journal-title":"arXiv 1811 12889"},{"key":"ref46","first-page":"1691","article-title":"Generative pretraining from pixels","author":"chen","year":"2020","journal-title":"Proc 37th Int Conf Mach Learn"},{"key":"ref45","doi-asserted-by":"publisher","DOI":"10.7551\/mitpress\/9780262033589.001.0001"},{"key":"ref48","article-title":"Recurrent environment simulators","author":"chiappa","year":"2017","journal-title":"Proc 5th Int Conf Learn Represent (ICLR)"},{"key":"ref47","article-title":"A simple framework for contrastive learning of visual representations","author":"chen","year":"2020","journal-title":"arXiv 2002 05709"},{"key":"ref42","article-title":"Multi-level cause-effect systems","author":"chalupka","year":"2015","journal-title":"arXiv 1512 07942"},{"key":"ref41","doi-asserted-by":"publisher","DOI":"10.1023\/A:1007379606734"},{"key":"ref44","article-title":"A compositional object-based approach to learning physical dynamics","author":"chang","year":"2017","journal-title":"Proc 5th Int Conf Learn Represent (ICLR)"},{"key":"ref43","article-title":"Fast conditional independence test for vector variables with large sample sizes","author":"chalupka","year":"2018","journal-title":"arXiv 1804 02747"},{"key":"ref73","first-page":"2052","article-title":"Off-policy deep reinforcement learning without exploration","author":"fujimoto","year":"2019","journal-title":"Proc Int Conf Mach Learn"},{"key":"ref72","article-title":"Autonomy of economic relations","author":"frisch","year":"1948"},{"key":"ref71","doi-asserted-by":"publisher","DOI":"10.1088\/0004-637X\/806\/2\/215"},{"key":"ref70","doi-asserted-by":"publisher","DOI":"10.1162\/neco.1991.3.2.194"},{"key":"ref76","article-title":"ImageNet-trained CNNs are biased towards texture; increasing shape bias improves accuracy and robustness","author":"geirhos","year":"2018","journal-title":"arXiv 1811 12231"},{"key":"ref77","first-page":"15740","article-title":"On the transfer of inductive bias from simulation to the real world: A new disentanglement dataset","author":"gondal","year":"2019","journal-title":"Proc Adv Neural Inf Process Syst"},{"key":"ref74","first-page":"489","article-title":"Kernel measures of conditional dependence","author":"fukumizu","year":"2008","journal-title":"Proc Adv Neural Inf Process Syst"},{"key":"ref75","doi-asserted-by":"publisher","DOI":"10.1007\/BF01531004"},{"key":"ref78","first-page":"2839","article-title":"Domain adaptation with conditional transferable components","author":"gong","year":"2016","journal-title":"Proc 33rd Int Conf Mach Learn"},{"key":"ref79","first-page":"269","article-title":"Causal discovery from temporally aggregated time series","author":"gong","year":"2017","journal-title":"Proc 33rd Conf Uncertainty Artif Intell (UAI)"},{"key":"ref60","article-title":"On robustness and transferability of convolutional neural networks","author":"djolonga","year":"2020","journal-title":"arXiv 2007 08558"},{"key":"ref62","article-title":"A framework for the quantitative evaluation of disentangled representations","author":"eastwood","year":"2018","journal-title":"Proc Int Conf Learn Represent"},{"key":"ref61","first-page":"132","article-title":"A permutation-based kernel conditional independence test","author":"doran","year":"2014","journal-title":"Proc Conf Uncertainty of Artificial Intelligence"},{"key":"ref63","first-page":"107","article-title":"Exact Bayesian structure learning from uncertain interventions","author":"eaton","year":"2007","journal-title":"Proc Artif Intell Statist"},{"key":"ref64","article-title":"Exploring the landscape of spatial robustness","author":"engstrom","year":"2017","journal-title":"arXiv 1712 02779"},{"key":"ref65","doi-asserted-by":"publisher","DOI":"10.1177\/1088868308316091"},{"key":"ref66","doi-asserted-by":"publisher","DOI":"10.1038\/s41591-018-0316-z"},{"key":"ref67","doi-asserted-by":"publisher","DOI":"10.1109\/18.243433"},{"key":"ref68","article-title":"Model-agnostic meta-learning for fast adaptation of deep networks","author":"finn","year":"2017","journal-title":"arXiv 1703 03400"},{"key":"ref69","article-title":"Counterfactual multi-agent policy gradients","author":"foerster","year":"2018","journal-title":"Proc 32nd AAAI Conf Artif Intell"},{"key":"ref197","article-title":"Spatially structured recurrent modules","author":"rahaman","year":"2021","journal-title":"Proc Int Conf Learn Represent"},{"key":"ref198","doi-asserted-by":"publisher","DOI":"10.1063\/1.3059791"},{"key":"ref199","doi-asserted-by":"publisher","DOI":"10.1023\/A:1009643025983"},{"key":"ref193","doi-asserted-by":"publisher","DOI":"10.1111\/rssb.12235"},{"key":"ref194","article-title":"An analysis of the adaptation speed of causal models","author":"priol","year":"2020","journal-title":"arXiv 2005 09136"},{"key":"ref195","article-title":"Failing loudly: An empirical study of methods for detecting dataset shift","author":"rabanser","year":"2018","journal-title":"arXiv 1810 11953"},{"key":"ref196","article-title":"Improving language understanding by generative pre-training","author":"radford","year":"2018"},{"key":"ref95","article-title":"Using videos to evaluate image model robustness","author":"gu","year":"2019","journal-title":"arXiv 1904 10076"},{"key":"ref190","first-page":"2009","article-title":"Causal discovery with continuous additive noise models","volume":"15","author":"peters","year":"2014","journal-title":"J Mach Learn Res"},{"key":"ref94","doi-asserted-by":"publisher","DOI":"10.1145\/280814.280816"},{"key":"ref191","doi-asserted-by":"publisher","DOI":"10.1073\/pnas.1905688116"},{"key":"ref93","article-title":"Bootstrap your own latent: A new approach to self-supervised learning","author":"grill","year":"2020","journal-title":"arXiv 2006 07733"},{"key":"ref192","doi-asserted-by":"publisher","DOI":"10.1080\/01621459.2018.1491403"},{"key":"ref92","first-page":"2075","article-title":"Kernel methods for measuring independence","volume":"6","author":"gretton","year":"2005","journal-title":"J Mach Learn Res"},{"key":"ref91","doi-asserted-by":"publisher","DOI":"10.1007\/11564089_7"},{"key":"ref90","article-title":"The incomplete Rosetta stone problem: Identifiability results for multi-view nonlinear ICA","author":"gresele","year":"2019","journal-title":"arXiv 1905 06642"},{"key":"ref98","first-page":"1","article-title":"Causality: Objectives and assessment","volume":"6","author":"guyon","year":"2010","journal-title":"Proc JMLR Workshop Conf"},{"key":"ref99","article-title":"World models","author":"ha","year":"2018","journal-title":"arXiv 1803 10122"},{"key":"ref96","article-title":"Towards deep neural network architectures robust to adversarial examples","author":"gu","year":"2014","journal-title":"arXiv 1412 5068 [cs]"},{"key":"ref97","article-title":"A survey of learning causality with data: Problems and methods","author":"guo","year":"2018","journal-title":"arXiv 1809 09337"},{"key":"ref82","article-title":"Evaluating reinforcement learning algorithms in observational health settings","author":"gottesman","year":"2018","journal-title":"arXiv 1805 12298"},{"key":"ref81","doi-asserted-by":"publisher","DOI":"10.1037\/0033-295X.111.1.3"},{"key":"ref84","article-title":"Object files and schemata: Factorizing declarative and procedural knowledge in dynamical systems","author":"goyal","year":"2020","journal-title":"arXiv 2006 16225"},{"key":"ref83","article-title":"Causal generative neural networks","author":"goudet","year":"2017","journal-title":"arXiv 1711 08936"},{"key":"ref80","article-title":"Explaining and harnessing adversarial examples","author":"goodfellow","year":"2014","journal-title":"arXiv 1412 6572"},{"key":"ref89","first-page":"13475","article-title":"Shaping belief states with generative environment models for RL","author":"gregor","year":"2019","journal-title":"Proc Adv Neural Inf Process Syst"},{"key":"ref85","article-title":"Recurrent independent mechanisms","author":"goyal","year":"2021","journal-title":"Proc Int Conf Learn Represent"},{"key":"ref86","doi-asserted-by":"publisher","DOI":"10.1109\/ICASSP.2013.6638947"},{"key":"ref87","first-page":"2424","article-title":"Multi-object representation learning with iterative variational inference","author":"greff","year":"2019","journal-title":"Proc Int Conf Mach Learn"},{"key":"ref88","article-title":"On the binding problem in artificial neural networks","author":"greff","year":"2020","journal-title":"arXiv 2012 05208"},{"key":"ref200","article-title":"Causally correct partial models for reinforcement learning","author":"rezende","year":"2020","journal-title":"arXiv 2002 02836"},{"key":"ref101","article-title":"Hidden Markov nonlinear ICA: Unsupervised learning from nonstationary time series","author":"h\u00e4lv\u00e4","year":"2020","journal-title":"arXiv 2006 12107"},{"key":"ref100","doi-asserted-by":"publisher","DOI":"10.2307\/1906935"},{"key":"ref209","article-title":"An overview of multi-task learning in deep neural networks","author":"ruder","year":"2017","journal-title":"ArXiv 1706 05098"},{"key":"ref203","doi-asserted-by":"publisher","DOI":"10.1037\/0022-3514.66.5.805"},{"key":"ref204","first-page":"1","article-title":"Invariant models for causal transfer learning","volume":"19","author":"rojas-carulla","year":"2018","journal-title":"J Mach Learn Res"},{"key":"ref201","doi-asserted-by":"publisher","DOI":"10.1038\/s41467-020-17419-7"},{"key":"ref202","first-page":"185","article-title":"Learning deep disentangled embeddings with the f-statistic loss","author":"ridgeway","year":"2018","journal-title":"Proc Adv Neural Inf Process Syst"},{"key":"ref207","first-page":"114","article-title":"From deterministic ODEs to dynamic structural causal models","author":"rubenstein","year":"2018","journal-title":"Proc 34th Conf Uncertainty Artif Intell (UAI)"},{"key":"ref208","first-page":"808","article-title":"Causal consistency of structural equation models","author":"rubenstein","year":"2017","journal-title":"Proc Thirty-Third Conf Uncertainty Artif Intell"},{"key":"ref205","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2019.01269"},{"key":"ref206","article-title":"Effects of degradations on deep neural network architectures","author":"roy","year":"2018","journal-title":"arXiv 1807 10108"},{"key":"ref211","article-title":"Learning to simulate complex physics with graph networks","author":"sanchez-gonzalez","year":"2020","journal-title":"arXiv 2002 09405"},{"key":"ref210","author":"russell","year":"2002","journal-title":"Artificial Intelligence A Modern Approach"},{"key":"ref212","first-page":"4967","article-title":"A simple neural network module for relational reasoning","author":"santoro","year":"2017","journal-title":"Proc Adv Neural Inf Process Syst"},{"key":"ref213","article-title":"Evolutionary principles in self-referential learning, or on learning how to learn: The meta-meta-… hook","author":"schmidhuber","year":"1987"},{"key":"ref214","doi-asserted-by":"publisher","DOI":"10.1109\/IJCNN.1991.170605"},{"key":"ref215","doi-asserted-by":"publisher","DOI":"10.1038\/518486a"},{"key":"ref216","article-title":"Causal learning, 2017","author":"sch\u00f6lkopf","year":"2017","journal-title":"Proc 34th Int Conf Mach Learn (ICML)"},{"key":"ref217","article-title":"Causality for machine learning","author":"sch\u00f6lkopf","year":"2019","journal-title":"arXiv 1911 10500"},{"key":"ref218","doi-asserted-by":"publisher","DOI":"10.1073\/pnas.1511656113"},{"key":"ref219","first-page":"1896","article-title":"Causal and statistical learning","volume":"13","author":"sch\u00f6lkopf","year":"2016","journal-title":"Oberwolfach Rep"},{"key":"ref220","first-page":"1255","article-title":"On causal and anticausal learning","author":"sch\u00f6lkopf","year":"2012","journal-title":"Proc 29th Int Conf Mach Learn (ICML)"},{"key":"ref222","article-title":"Towards the first adversarially robust neural network model on MNIST","author":"schott","year":"2019","journal-title":"Proc Int Conf Learn Represent"},{"key":"ref221","author":"sch\u00f6lkopf","year":"2002","journal-title":"Learning with kernels"},{"key":"ref229","first-page":"2003","article-title":"A linear non-Gaussian acyclic model for causal discovery","volume":"7","author":"shimizu","year":"2006","journal-title":"J Mach Learn Res"},{"key":"ref228","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2019.00841"},{"key":"ref227","article-title":"Do image classifiers generalize across time","author":"shankar","year":"2019","journal-title":"arXiv 1906 02168"},{"key":"ref226","first-page":"285","article-title":"Telling cause from effect in deterministic linear dynamical systems","author":"shajarisales","year":"2015","journal-title":"Proc 32nd Int Conf Mach Learn (ICML)"},{"key":"ref225","article-title":"The hardness of conditional independence testing and the generalised covariance measure","author":"shah","year":"2018","journal-title":"arXiv 1804 07203"},{"key":"ref224","first-page":"1697","article-title":"Reliable decision support using counterfactual models","author":"schulam","year":"2017","journal-title":"Proc Adv Neural Inf Process Syst"},{"key":"ref223","article-title":"Mastering Atari, go, chess and Shogi by planning with a learned model","author":"schrittwieser","year":"2019","journal-title":"arXiv 1911 08265"},{"key":"ref127","doi-asserted-by":"publisher","DOI":"10.1613\/jair.301"},{"key":"ref126","first-page":"2250","article-title":"Detecting non-causal artifacts in multivariate linear regression models","author":"janzing","year":"2018","journal-title":"Proc 35th Int Conf Mach Learn (ICML)"},{"key":"ref125","first-page":"1923","article-title":"Semi-supervised interpolation in an anticausal learning scenario","volume":"16","author":"janzing","year":"2015","journal-title":"J Mach Learn Res"},{"key":"ref124","doi-asserted-by":"publisher","DOI":"10.1109\/TIT.2010.2060095"},{"key":"ref129","doi-asserted-by":"publisher","DOI":"10.1109\/BIOSIG.2016.7736924"},{"key":"ref128","author":"kahneman","year":"2011","journal-title":"Thinking Fast and Slow"},{"key":"ref130","article-title":"Algorithmic recourse under imperfect causal knowledge: A probabilistic approach","author":"karimi","year":"2020","journal-title":"arXiv 2006 06831"},{"key":"ref133","article-title":"Generalization in anti-causal learning","author":"kilbertus","year":"2018","journal-title":"arXiv 1812 00524"},{"key":"ref134","first-page":"656","article-title":"Avoiding discrimination through causal reasoning","author":"kilbertus","year":"2017","journal-title":"Proc Adv Neural Inf Process Syst"},{"key":"ref131","article-title":"Learning neural causal models from unknown interventions","author":"ke","year":"2019","journal-title":"arXiv 1910 01075"},{"key":"ref132","article-title":"Optimal decision making under strategic behavior","author":"tsirtsis","year":"2019","journal-title":"arXiv 1905 09239"},{"key":"ref232","doi-asserted-by":"crossref","first-page":"484","DOI":"10.1038\/nature16961","article-title":"Mastering the game of go with deep neural networks and tree search","volume":"529","author":"silver","year":"2016","journal-title":"Nature"},{"key":"ref233","first-page":"895","article-title":"Tangent prop—A formalism for specifying selected invariances in an adaptive network","volume":"4","author":"simard","year":"1992","journal-title":"Advances in neural information processing systems"},{"key":"ref230","article-title":"Weakly supervised disentanglement with guarantees","author":"shu","year":"2019","journal-title":"arXiv 1910 09772"},{"key":"ref231","first-page":"3191","article-title":"The predictron: End-to-end learning and planning","author":"silver","year":"2017","journal-title":"Proc Int Conf Mach Learn (PMLR)"},{"key":"ref239","author":"steinwart","year":"2008","journal-title":"Support Vector Machines"},{"key":"ref238","author":"spohn","year":"1978","journal-title":"Grundlagen der Entscheidungstheorie"},{"key":"ref235","first-page":"49","article-title":"Causal ordering and identifiability","author":"simon","year":"1953","journal-title":"Studies in Econometric Method"},{"key":"ref234","doi-asserted-by":"publisher","DOI":"10.1109\/ICDAR.2003.1227801"},{"key":"ref237","author":"spirtes","year":"2000","journal-title":"Causation Prediction and Search"},{"key":"ref236","doi-asserted-by":"publisher","DOI":"10.1207\/s15516709cog1401_3"},{"key":"ref136","first-page":"2688","article-title":"Neural relational inference for interacting systems","author":"kipf","year":"2018","journal-title":"Proc Int Conf Mach Learn"},{"key":"ref135","first-page":"2649","article-title":"Disentangling by factorising","author":"kim","year":"2018","journal-title":"Proc Int Conf Mach Learn"},{"key":"ref138","first-page":"8606","article-title":"Sequential attend, infer, repeat: Generative modelling of moving objects","volume":"31","author":"kosiorek","year":"2018","journal-title":"Proc Adv Neural Inf Process Syst"},{"key":"ref137","article-title":"Big transfer (BiT): General visual representation learning","author":"kolesnikov","year":"2019","journal-title":"arXiv 1912 11370"},{"key":"ref139","first-page":"478","article-title":"Consistency of causal inference under the additive noise model","author":"kpotufe","year":"2014","journal-title":"Proc 31st Int Conf Mach Learn"},{"key":"ref140","first-page":"1097","article-title":"ImageNet classification with deep convolutional neural networks","author":"krizhevsky","year":"2012","journal-title":"Proc Adv Neural Inf Process Syst"},{"key":"ref141","first-page":"10723","article-title":"Unsupervised learning of object keypoints for perception and control","author":"kulkarni","year":"2019","journal-title":"Proc Adv Neural Inf Process Syst"},{"key":"ref142","first-page":"4066","article-title":"Counterfactual fairness","author":"kusner","year":"2017","journal-title":"Advances in neural information processing systems"},{"key":"ref143","doi-asserted-by":"publisher","DOI":"10.1145\/2816795.2818129"},{"key":"ref144","doi-asserted-by":"publisher","DOI":"10.1017\/S0140525X16001837"},{"key":"ref2","article-title":"Solving Rubik’s cube with a robot hand","year":"2019","journal-title":"arXiv 1910 07113"},{"key":"ref145","doi-asserted-by":"publisher","DOI":"10.1007\/BF02251257"},{"key":"ref1","article-title":"Causalworld: A robotic manipulation benchmark for causal structure and transfer learning","author":"ahmed","year":"2021","journal-title":"Proc Int Conf Learn Represent"},{"key":"ref241","article-title":"Counterfactual multi-agent reinforcement learning with graph convolution communication","author":"su","year":"2020","journal-title":"arXiv 2004 00470"},{"key":"ref242","article-title":"Counterfactual normalization: Proactively addressing dataset shift and improving reliability using causal mechanisms","author":"subbaswamy","year":"2018","journal-title":"arXiv 1808 03253"},{"key":"ref243","article-title":"Preventing failures due to dataset shift: Learning predictive models that transport","author":"subbaswamy","year":"2018","journal-title":"arXiv 1812 04597"},{"key":"ref244","article-title":"Stochastic prediction of multi-agent interactions from partial observations","author":"sun","year":"2019","journal-title":"arXiv 1902 09641"},{"key":"ref240","first-page":"464","article-title":"Causal Markov condition for submodular information measures","author":"steudel","year":"2010","journal-title":"Proc 23rd Annu Conf Learn Theory (COLT)"},{"key":"ref248","volume":"135","author":"sutton","year":"1998","journal-title":"Introduction to Reinforcement Learning"},{"key":"ref247","first-page":"6056","article-title":"Robustly disentangled causal mechanisms: Validating deep representations for interventional robustness","author":"suter","year":"2019","journal-title":"Proc Int Conf Mach Learn (PMLR)"},{"key":"ref246","first-page":"1","article-title":"Causal inference by choosing graphs with most plausible Markov kernels","author":"sun","year":"2006","journal-title":"Proc 9th Int Symp Artif Intell Math"},{"key":"ref245","doi-asserted-by":"publisher","DOI":"10.1109\/ICCV.2017.97"},{"key":"ref249","article-title":"Intriguing properties of neural networks","author":"szegedy","year":"2013","journal-title":"arXiv 1312 6199"},{"key":"ref109","doi-asserted-by":"publisher","DOI":"10.1126\/scitranslmed.aab3719"},{"key":"ref108","doi-asserted-by":"publisher","DOI":"10.1515\/9781400873296"},{"key":"ref107","article-title":"Benchmarking neural network robustness to common corruptions and perturbations","author":"hendrycks","year":"2019","journal-title":"arXiv 1903 12261"},{"key":"ref106","article-title":"Invariant causal prediction for nonlinear models","author":"heinze-deml","year":"2017","journal-title":"arXiv 1706 08576"},{"key":"ref105","article-title":"Conditional variance penalties and domain shift robustness","author":"heinze-deml","year":"2017","journal-title":"arXiv 1710 11469"},{"key":"ref104","doi-asserted-by":"publisher","DOI":"10.1073\/pnas.1821458116"},{"key":"ref103","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2016.90"},{"key":"ref102","first-page":"9729","article-title":"Momentum contrast for unsupervised visual representation learning","author":"he","year":"2020","journal-title":"Proc IEEE Conf Comput Vis Pattern Recognit (CVPR)"},{"key":"ref111","doi-asserted-by":"publisher","DOI":"10.1057\/978-1-349-95121-5_2227-1"},{"key":"ref112","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/P18-1031"},{"key":"ref110","article-title":"Beta-VAE: Learning basic visual concepts with a constrained variational framework","author":"higgins","year":"2016","journal-title":"Proc Int Conf Learn Represent"},{"key":"ref250","doi-asserted-by":"publisher","DOI":"10.1126\/science.1196404"},{"key":"ref251","first-page":"512","article-title":"Causal discovery from changes","author":"tian","year":"2001","journal-title":"Proc 17th Annual Conf Uncertainty Artif Intell (UAI)"},{"key":"ref254","doi-asserted-by":"publisher","DOI":"10.1038\/s41550-019-0878-9"},{"key":"ref255","article-title":"Relational neural expectation maximization: Unsupervised discovery of objects and their interactions","author":"van steenkiste","year":"2018","journal-title":"Proc 6th Proc Int Conf Learn Represent (ICLR)"},{"key":"ref252","article-title":"Is independence all you need? On the generalization of representations learned from correlated data","author":"tr\u00e4uble","year":"2020","journal-title":"arXiv 2006 07886"},{"key":"ref253","first-page":"13806","article-title":"Self-supervised learning of video-induced visual invariances","author":"tschannen","year":"2020","journal-title":"Proc IEEE Conf Comput Vis Pattern Recognit (CVPR)"},{"key":"ref257","author":"vapnik","year":"1998","journal-title":"Statistical Learning Theory"},{"key":"ref256","first-page":"14178","article-title":"Are disentangled representations helpful for abstract visual reasoning","author":"van steenkiste","year":"2019","journal-title":"Proc Adv Neural Inf Process Syst"},{"key":"ref10","first-page":"38","article-title":"Document image defect models","author":"baird","year":"1990","journal-title":"Proceedings of the IAPR Workshop on Syntactic and Structural Pattern Recognition"},{"key":"ref259","article-title":"On the fairness of causal algorithmic recourse","author":"von k\u00fcgelgen","year":"2020","journal-title":"arXiv 2010 06529"},{"key":"ref11","first-page":"464","article-title":"Structured agents for physical construction","author":"bapst","year":"2019","journal-title":"Proc Int Conf Mach Learn"},{"key":"ref258","doi-asserted-by":"publisher","DOI":"10.1038\/s41586-019-1724-z"},{"key":"ref12","first-page":"9448","article-title":"Objectnet: A large-scale bias-controlled dataset for pushing the limits of object recognition models","author":"barbu","year":"2019","journal-title":"Proc Adv Neural Inf Process Syst"},{"key":"ref13","first-page":"1342","article-title":"Bandits with unobserved confounders: A causal approach","author":"bareinboim","year":"2015","journal-title":"Proc Adv Neural Inf Process Syst"},{"key":"ref14","first-page":"280","article-title":"Transportability from multiple environments with limited experiments: Completeness results","author":"bareinboim","year":"2014","journal-title":"Proc Adv Neural Inf Process Syst"},{"key":"ref15","first-page":"4502","article-title":"Interaction networks for learning about objects, relations and physics","author":"battaglia","year":"2016","journal-title":"Proc Adv Neural Inf Process Syst"},{"key":"ref118","doi-asserted-by":"publisher","DOI":"10.1017\/CBO9781139025751"},{"key":"ref16","article-title":"Relational inductive biases, deep learning, and graph networks","author":"battaglia","year":"2018","journal-title":"arXiv 1806 01261"},{"key":"ref117","doi-asserted-by":"publisher","DOI":"10.1016\/S0893-6080(98)00140-3"},{"key":"ref17","doi-asserted-by":"publisher","DOI":"10.1073\/pnas.1306572110"},{"key":"ref18","first-page":"2043","article-title":"The arrow of time in multivariate time series","volume":"48","author":"bauer","year":"2016","journal-title":"Proc 33rd Int Conf Mach Learn"},{"key":"ref119","first-page":"12704","article-title":"Causal regularization","author":"janzing","year":"2019","journal-title":"Proc Adv Neural Inf Process Syst"},{"key":"ref19","doi-asserted-by":"publisher","DOI":"10.1145\/3313831.3376718"},{"key":"ref114","first-page":"1","article-title":"Causal discovery from heterogeneous\/nonstationary data","volume":"21","author":"huang","year":"2020","journal-title":"J Mach Learn Res"},{"key":"ref113","first-page":"689","article-title":"Nonlinear causal discovery with additive noise models","author":"hoyer","year":"2009","journal-title":"Proc Adv Neural Inf Process Syst (NIPS)"},{"key":"ref116","first-page":"460","article-title":"Nonlinear ICA of temporally dependent stationary sources","author":"hyvarinen","year":"2017","journal-title":"Mach Learn Res"},{"key":"ref115","doi-asserted-by":"publisher","DOI":"10.1109\/ICDM.2017.114"},{"key":"ref120","doi-asserted-by":"publisher","DOI":"10.1088\/1367-2630\/18\/9\/093052"},{"key":"ref121","first-page":"479","article-title":"Telling cause from effect based on high-dimensional observations","author":"janzing","year":"2010","journal-title":"Proc 27th Int Conf Mach Learn"},{"key":"ref122","doi-asserted-by":"publisher","DOI":"10.1016\/j.artint.2012.01.002"},{"key":"ref123","first-page":"249","article-title":"Identifying confounders using additive noise models","author":"janzing","year":"2009","journal-title":"Proc 25th Annu Conf Uncertainty Artif Intell (UAI)"},{"key":"ref260","article-title":"Simpson’s paradox in COVID-19 case fatality rates: A mediation analysis of age-related causal effects","author":"von k\u00fcgelgen","year":"2020","journal-title":"arXiv 2005 07180"},{"key":"ref261","first-page":"1","article-title":"Semi-supervised learning, causality and the conditional cluster assumption","author":"von k\u00fcgelgen","year":"2020","journal-title":"Proc Conf Uncertainty Artif Intell (UAI)"},{"key":"ref262","article-title":"Towards causal generative scene models via competition of experts","author":"von k\u00fcgelgen","year":"2020","journal-title":"arXiv 2004 12906"},{"key":"ref263","article-title":"Learning robust representations by projecting superficial statistics out","author":"wang","year":"2019","journal-title":"arXiv 1903 06256"},{"key":"ref264","article-title":"COBRA: Data-efficient model-based RL through unsupervised object discovery and curiosity-driven exploration","author":"watters","year":"2019","journal-title":"arXiv 1905 09275"},{"key":"ref265","first-page":"4539","article-title":"Visual interaction networks: Learning a physics simulator from video","author":"watters","year":"2017","journal-title":"Proc Adv Neural Inf Process Syst"},{"key":"ref266","article-title":"Pragmatism and variable transformations in causal modelling","author":"weichwald","year":"2019"}],"container-title":["Proceedings of the IEEE"],"original-title":[],"link":[{"URL":"http:\/\/xplorestaging.ieee.org\/ielx7\/5\/9420072\/09363924.pdf?arnumber=9363924","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2024,8,24]],"date-time":"2024-08-24T19:19:11Z","timestamp":1724527151000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/9363924\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2021,5]]},"references-count":282,"journal-issue":{"issue":"5"},"URL":"https:\/\/doi.org\/10.1109\/jproc.2021.3058954","relation":{},"ISSN":["0018-9219","1558-2256"],"issn-type":[{"value":"0018-9219","type":"print"},{"value":"1558-2256","type":"electronic"}],"subject":[],"published":{"date-parts":[[2021,5]]}}}