[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-117840-en":3,"doc-seo-117840-105":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},117840,962075006959,"Anda","https://ap-avatar.wpscdn.com/avatar/e0002397efbe92a78e?_k=1776741047341049297",8,"Research & Report","Adaptive debiased machine learning using data-driven model selection techniques","Debiased machine learning estimators for nonparametric inference of smooth functionals can suffer from excessive variability and instability. Practitioners may switch to parametric or semiparametric models, but misspecification then induces bias. Adaptive Debiased Machine Learning (ADML) combines data-driven model selection with debiasing to build asymptotically linear, adaptive, and superefficient estimators for pathwise differentiable functionals. ADML learns the oracle submodel from data, avoiding local asymptotic penalties for data-driven selection, and applies broadly to average treatment effect estimation in adaptive partially linear regression models.","arXiv :2307 . 12544v1 [ stat .ME] 24 Jul 2023  \nAdaptive debiased machine learning using data-driven model selection techniques  \nLars van der Laan, Marco Carone, Alex Luedtke, Mark van der Laan  \nJuly 25, 2023  \nAbstract  \nDebiased machine learning estimators for nonparametric inference of smooth functionals of the data-generating distribution can su􀀋er from excessive variability and instability. For this reason, practitioners may resort to simpler models based on parametric or semiparametric assumptions. However, such simplifying assumptions may fail to hold, and estimates may then be biased due to model misspeci􀀌cation. To address this problem, we propose Adaptive Debiased Machine Learning (ADML), a nonparametric framework that combines data-driven modelselection and debiased machine learning techniques to construct asymptotically linear, adaptive, and supere􀀎cient estimators for pathwise di􀀋erentiable functionals. By learning model structure directly from data, ADML avoids the bias introduced by model misspeci􀀌cation and remains free from the restrictions of parametric and semiparametric models. While they may exhibit irregular behavior for the target parameter in a nonparametric statistical model, we demonstrate that ADML estimators provides regular and locally uniformly valid inference for a projection-based oracle parameter. Importantly, this oracle parameter agrees with the original target parameter for distributions within an unknown but correctly speci􀀌ed oracle statistical submodel that is learned from the data. This 􀀌nding implies that there is no penalty, in a local asymptotic sense, for conducting data-driven model selection compared to having prior knowledge of the oracle submodel and oracle parameter. To demonstrate the practical applicability of our theory, we provide a broad class of ADML estimators for estimating the average treatment e􀀋ect in adaptive partially linear regression models.  \nKeywords | Adaptive debiased machine learning, causal inference, supere􀀎cient, model selection, selective inference.  \n1. Introduction  \nFor many scienti􀀌c applications, including treatment e􀀋ect estimation and policy learning, it is critical to infer real-valued summaries (i.e., functionals) of probability distributions. For this purpose, several debiased machine learning frameworks are available, including one-step estimation (Pfanzagl and Wefelmeyer, 1985; Bickel et al., 1993), estimating equations and double-machine learning (Robins et al., 1995, 1994; van der Laan and Robins, 2003; Chernozhukov et al., 2018a), targeted maximum likelihood estimation (Laan and Rubin, 2006; van der Laan and Rose, 2011), and sieve-based plug-in estimation (Shen, 1997; Chen, 2007; Chen and Liao, 2014; Qiu et al., 2020; van der Laan and Rose, 2021; van der Laan et al., 2022) . These frameworks typically involve two stages: preliminary estimation, wherein 􀀍exible machine learning techniques are used to estimate the data-generating distribution; and debiasing, which facilitates valid uncertainty assessment based  \non a prespeci􀀌ed statistical model. When the model is correctly speci􀀌ed, these methods yield parametricrate consistent, regular, and asymptotically linear estimators that are e􀀎cient among the class of all regular estimators (Bickel et al., 1993; van der Vaart, 2000) . Additionally, such e􀀎cient estimators are locally asymptotically minimax among all estimators, including irregular estimators, with respect to the statistical model (van der Vaart, 2000) . In other words, asymptotically, they minimize the maximum mean square estimation error over all local perturbations of the data-generating distribution that fall within the statistical model, that is, over all local alternatives.  \nWhile debiasing approaches have proven e􀀋ective in generating e􀀎cient and locally asymptotically minimax estimators, they do possess a notable limitation: the debiasing step and uncertainty quanti􀀌cation necessitate a priori speci􀀌cation of a correct sta","cbCaiciSAnSAFujy","https://ap.wps.com/l/cbCaiciSAnSAFujy","pdf",983369,1,46,"English","en",105,"# Introduction\n## Debiased machine learning and efficiency limits\n## Oracle submodels and adaptivity challenge\n## Model misspecification bias vs variance trade-off\n## Proposed ADML framework and goals","[{\"question\":\"Why can debiased machine learning estimators be unstable in nonparametric inference?\",\"answer\":\"They can exhibit excessive variability and instability for smooth functionals when uncertainty quantification relies on a prespecified model.\"},{\"question\":\"What is the main limitation of standard debiasing methods?\",\"answer\":\"They require an a priori correct statistical model, which prevents adaptivity to the true (possibly structured) data-generating distribution.\"},{\"question\":\"How does ADML address data-driven model selection without losing validity?\",\"answer\":\"ADML combines adaptive model selection with debiasing to produce asymptotically linear, adaptive, and superefficient estimators, learning an oracle submodel so that local asymptotic inference remains valid.\"}]","Adaptive debiased machine learning using data-driven model selection techniques | PDF",1785679939,116,{"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},"adaptive-debiased-machine-learning-using-data-driven-model-selection-techniques","",{"@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/adaptive-debiased-machine-learning-using-data-driven-model-selection-techniques/117840/",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},"Why can debiased machine learning estimators be unstable in nonparametric inference?","Question",{"text":75,"@type":76},"They can exhibit excessive variability and instability for smooth functionals when uncertainty quantification relies on a prespecified model.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"What is the main limitation of standard debiasing methods?",{"text":80,"@type":76},"They require an a priori correct statistical model, which prevents adaptivity to the true (possibly structured) data-generating distribution.",{"name":82,"@type":73,"acceptedAnswer":83},"How does ADML address data-driven model selection without losing validity?",{"text":84,"@type":76},"ADML combines adaptive model selection with debiasing to produce asymptotically linear, adaptive, and superefficient estimators, learning an oracle submodel so that local asymptotic inference remains valid.","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"]