[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-117223-en":3,"doc-seo-117223-105":30,"detail-sidebar-cat-0-en-105":91},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":4,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":27,"seo_description":14,"update_tm":28,"read_time":29},117223,2336464648322,"Aria","https://ap-avatar.wpscdn.com/avatar/2200025388227c56fec?_k=1778556882303663488",8,"Research & Report","Automatic debiased machine learning for covariate shifts","This paper addresses bias that arises when machine learning estimates parameters under covariate shift—when the distribution of input features differs between training and deployment. Regularization and model selection can make many parameter estimates biased. An automatic debiased machine learning approach is proposed that corrects this bias without requiring an explicit bias-correction formula. The method debiases estimators of policy and causal parameters using only the parameter of interest and relies on debiased machine learning techniques. As sample size grows, the estimator is shown to be asymptotically normal, and is evaluated on a regression setting via Monte Carlo simulations.","arXiv :2307 .04527v3 [ stat .ME] 19 Apr 2024  \nAutomatic debiased machine learning for  \ncovariate shifts  \nVictor Chernozhukov 1 , Michael Newey2 , Whitney K. Newey 1 , Rahul Singh3 ,  \nand Vasilis Syrgkanis4  \n1 Department of Economics, Massachusetts Institute of Technology, Cambridge, MA 02142, USA  \n2 Lincoln Laboratory, Massachusetts Institute of Technology, Lexington, MA 02421, USA  \n3 Society of Fellows and Department of Economics, Harvard University, Cambridge, MA 02138, USA  \n4 Department of Management Science and Engineering, Stanford University, Stanford, CA 94305, USA  \nOriginal draft: July 2023 . This draft: March 2024 .  \nAbstract  \nIn this paper we address the problem of bias in the machine learning of parameters following covariate shifts. Covariate shift occurs when the distribution of input features changes between the training and deployment stages. Regularization and model selection associated with machine learning biases many parameter estimates. In this paper, we propose an automatic debiased machine learning approach to correct for this bias under covariate shift. The proposed approach leverages state-of-theart debiased machine learning techniques to debias estimators of policy and causal parameters when covariate shift is present. The debiasing is automatic in only relying on the parameter of interest and not requiring the form of the bias. We show that our estimator is asymptotically normal as the sample size grows. Finally, we demonstrate the proposed method on a regression problem using a Monte Carlo simulation.  \nKeywords: Covariate shift; debiased machine learning; semiparametric inference; Gaussian approximation.  \n1 Introduction  \nIn applications, machine learners trained on one data set may be used to estimate parameters of interest in another data set with a different distribution of predictors. For example, the training data could be a sub-population of a larger population, or the training and estimation could take place at different times where the distribution of predictors varies between times. A case we consider in this work is a neural network trained to predict on one data set and then used to learn average outcomes from previously unseen data on predictor variables.  \nAnother case, from causal statistics, is a Lasso regression trained on outcome, treatment, and covariate data that is used to estimate counterfactual averages in another data set with a different distribution of covariates. This is an important case of distribution shift that is known as covariate shift (Shimodaira, 2000; Quiñonero-Candela et al., 2008; Pathak et al., 2022; Ma et al., 2023) . Such covariate shifts are of interest in a wide variety of settings, including estimation of counterfactual averages and causal effects with shifted covariates (Hotz et al., 2005; Pearl and Bareinboim, 2011; Singh et al., 2023) . Additionally, covariate shifts are interesting for classification where the training data may differ from the field data (Bahng et al., 2022; Koh et al., 2021) . There are many important parameters that depend on covariate shifts, including average outcomes or average potential outcomes learned from shifted covariate data.  \nThis paper concerns machine learning of parameters of interest in field data that depend on regressions in training data. An important problem with estimation is the bias that can result from regularization and/or model selection in machine learning on training data. In this paper we address this problem by giving automatic debiased machine learners of parameters of interest. The debiasing is automatic in only requiring the object of interest and in not requiring a full theoretical formula for the bias correction.  \nThe debiased estimators given are obtained by plugging a training data regression into the formula of interest in the field data and adding a debiasing term. The debiasing term consists of an average product in the training data of a debiasing function and regression residuals","cbCaiqjBnS43ie40","https://ap.wps.com/l/cbCaiqjBnS43ie40","pdf",521129,1,23,"English","en",105,"# Introduction\n## Parameters of interest under covariate shift\n# Derivation\n## Parameters of interest and debiased estimating equations","[{\"question\":\"What problem does the paper study under covariate shift?\",\"answer\":\"It studies how bias can be introduced into parameter estimates by regularization and model selection when feature distributions differ between training and deployment.\"},{\"question\":\"How does the proposed approach perform bias correction?\",\"answer\":\"It builds debiased estimators by plugging a training regression into the field-data parameter formula and adding a debiasing term based on an estimated debiasing function and training residuals.\"},{\"question\":\"What theoretical and empirical results are provided?\",\"answer\":\"The paper shows the estimator is asymptotically normal as sample size increases and demonstrates the method using a regression problem evaluated with Monte Carlo simulation.\"}]","Automatic debiased machine learning for covariate shifts | PDF",1785674483,58,{"code":4,"msg":31,"data":32},"ok",{"site_id":24,"language":23,"slug":33,"title":13,"keywords":34,"description":14,"schema_data":35,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":28},"automatic-debiased-machine-learning-for-covariate-shifts","",{"@graph":36,"@context":85},[37,54,68],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,48,51],{"item":41,"name":42,"@type":43,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":47},"https://docshare.wps.com/document/","Document",2,{"item":49,"name":12,"@type":43,"position":50},"https://docshare.wps.com/document/research-report/",3,{"item":52,"name":13,"@type":43,"position":53},"https://docshare.wps.com/document/automatic-debiased-machine-learning-for-covariate-shifts/117223/",4,{"url":52,"name":13,"@type":55,"author":56,"headline":13,"publisher":58,"fileFormat":61,"inLanguage":23,"description":14,"dateModified":62,"datePublished":62,"encodingFormat":61,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":57},"Person",{"url":41,"name":59,"@type":60},"DocShare","Organization","application/pdf","2026-08-02",true,{"@type":65,"interactionType":66,"userInteractionCount":4},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"What problem does the paper study under covariate shift?","Question",{"text":75,"@type":76},"It studies how bias can be introduced into parameter estimates by regularization and model selection when feature distributions differ between training and deployment.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"How does the proposed approach perform bias correction?",{"text":80,"@type":76},"It builds debiased estimators by plugging a training regression into the field-data parameter formula and adding a debiasing term based on an estimated debiasing function and training residuals.",{"name":82,"@type":73,"acceptedAnswer":83},"What theoretical and empirical results are provided?",{"text":84,"@type":76},"The paper shows the estimator is asymptotically normal as sample size increases and demonstrates the method using a regression problem evaluated with Monte Carlo simulation.","https://schema.org",{"og:url":52,"og:type":87,"og:title":13,"og:site_name":59,"og:description":14},"article",{"robots":89,"canonical":52},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":92},[93,97,101,105,110,115,120,123,128,131,135],{"id":20,"doc_module":4,"doc_module_name":46,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":47,"doc_module":4,"doc_module_name":46,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":53,"doc_module":4,"doc_module_name":46,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":106,"doc_module":4,"doc_module_name":46,"category_name":107,"show_sort_weight":108,"slug":109},5,"Comic",60,"comic",{"id":111,"doc_module":4,"doc_module_name":46,"category_name":112,"show_sort_weight":113,"slug":114},6,"Technology",50,"technology",{"id":116,"doc_module":4,"doc_module_name":46,"category_name":117,"show_sort_weight":118,"slug":119},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":121,"slug":122},30,"research-report",{"id":124,"doc_module":4,"doc_module_name":46,"category_name":125,"show_sort_weight":126,"slug":127},9,"Religion & Spirituality",20,"religion-spirituality",{"id":126,"doc_module":4,"doc_module_name":46,"category_name":129,"show_sort_weight":126,"slug":130},"World Cup","world-cup",{"id":132,"doc_module":4,"doc_module_name":46,"category_name":133,"show_sort_weight":132,"slug":134},10,"Lifestyle","lifestyle",{"id":136,"doc_module":4,"doc_module_name":46,"category_name":137,"show_sort_weight":106,"slug":138},19,"General","general"]