[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-118236-en":3,"doc-seo-118236-105":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},118236,1099514068035,"Ezra","https://ap-avatar.wpscdn.com/davatar_276721f389ce27ea32af1340a28f341c",8,"Research & Report","SYNERGIZING CAUSAL INFERENCE AND MACHINE LEARNING FOR ACTIONABLE INFERENCE","Rapid advances in storage and data-processing technologies have enabled collecting and analyzing multimodal data in domains such as healthcare, social media, and genomics, driving demand for data-driven inference that automates decisions or supports them. Supervised machine learning and causal inference both offer inference, yet supervised methods lack explainability and may fail to generalize across domains, while causal methods rely on additional assumptions and can require high-dimensional nuisance models, limiting actionability. This dissertation presents integrated methods where causal inference and machine learning jointly address these limitations through robust shift-intervention inference, equilibrium-dynamics causal effects, causal explanations for black-box behavior, tighter bounds for probabilities of causation, and mediation estimation with flexible models.","SYNERGIZING CAUSAL INFERENCE AND MACHINE LEARNING  \nFOR ACTIONABLE INFERENCE  \nby  \nNumair Sani  \nA dissertation submitted to Johns Hopkins University in conformity with the requirements for the degree of Doctor of Philosophy  \nBaltimore, Maryland  \nMay, 2024  \n© 2024 Numair Sani  \nAll rights reserved  \nAbstract  \nThe rapid development of storage systems and data-processing technologies in recent years has enabled the collection and analysis of various modalities of data generated in settings such as healthcare, social media, and genomics. This has led to a heightened interest in applying data-driven inference algorithms to automate decision-making or provide decision support. Supervised machine learning and causal inference are two important classes of data-driven inference algorithms that have received considerable attention. However, both have shortcomings that hamper their ability to provide actionable inference. The lack of explainability of supervised machine learning algorithms reduces trust in their predictions, thereby decreasing their chances of real-world deployment. Next, algorithms are often deployed in settings different from those in which they were trained, leading to inference results that may not generalize to other domains. Additionally, not every inference task involves prediction, for example, reasoning about the effect of interventions or attributing causes to the observed effects. Although causal approaches have shown promise in remedying each of these shortcomings, they make additional assumptions and often require modeling high-dimensional nuisance parameters, thereby diminishing their ability to provide actionable inference.  \nIn this dissertation, I present methods that illustrate the synergy between supervised machine learning and causal inference and how they can address each other’s limitations. To demonstrate how causal inference can address some shortcomings of machine learning algorithms, I first provide inference algorithms for shift interventions that have applications in robust machine learning, healthcare, and social sciences. I  \nthen provide an inference framework that utilizes causal inference to estimate causal effects in systems with equilibrium dynamics. Next, I present a technique for obtaining causal explanations for the behavior of black-box algorithms by utilizing causal discovery algorithms. Subsequently, I present approaches to tighten the bounds on the Probabilities of Causation, which are useful for causal attribution and policy evaluation. To demonstrate how machine learning can help causal inference, I present an approach for estimating mediation effects in the presence of real-valued treatments using flexible machine learning algorithms. This dissertation concludes by providing closing remarks and directions for future research.  \nThesis Readers  \nDr. Ilya Shpitser (Primary Advisor) John C. Malone Associate Professor Department of Computer Science  \nJohns Hopkins University  \nDr. Anqi Liu  \nAssistant Professor  \nDepartment of Computer Science  \nJohns Hopkins University  \nDr. Atalanti A. Mastakouri  \nApplied Scientist  \nCausality Lab, Amazon Research, Tübingen  \nTo my mentors, past and present, I would not have made it this far without your love  \nand guidance .  \nAcknowledgements  \nIt takes a village to raise a child, and I am immensely thankful to the people who have supported me along mine. First, to my late father, who tirelessly drove me to my squash practices every day on his motorcycle, which allowed me to obtain a scholarship to pursue my undergraduate studies in the United States. To my mother, who cared for me through the stressful period of final exams and college applications, I apologize for all the missed dinners, birthdays, and anniversaries. Next, I am grateful to the Rochester Squash coaching staff and men’s team, for challenging me everyday and teaching me the value of consistency and believing in myself. Fast forward to my time at Johns Hopkins, I am grate","cbCairGjUBC9a6aX","https://ap.wps.com/l/cbCairGjUBC9a6aX","pdf",4139016,1,347,"English","en",105,"# Abstract\n# Thesis Readers\n# Dedication\n# Acknowledgements\n# Contents\n## List of Figures\n# Chapter 1 Introduction","[{\"question\":\"Why do supervised machine learning and causal inference each struggle to provide actionable inference?\",\"answer\":\"Supervised machine learning can be difficult to explain and may not generalize when deployed in settings different from training. Causal inference can require additional assumptions and may involve modeling high-dimensional nuisance parameters, reducing actionability.\"},{\"question\":\"What main synergy does the dissertation propose between causal inference and machine learning?\",\"answer\":\"It presents methods showing how causal inference can mitigate limitations of machine learning and how machine learning can strengthen causal inference. Together, they aim to produce more trustworthy and broadly useful inference.\"},{\"question\":\"What types of methods are introduced to improve causal reasoning and attribution?\",\"answer\":\"The dissertation includes inference algorithms for shift interventions, a framework for causal effects in systems with equilibrium dynamics, causal explanations for black-box behavior using causal discovery, and techniques that tighten bounds on probabilities of causation.\"}]","SYNERGIZING CAUSAL INFERENCE AND MACHINE LEARNING FOR ACTIONABLE INFERENCE | PDF",1785682536,874,{"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},"synergizing-causal-inference-and-machine-learning-for-actionable-inference","",{"@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/synergizing-causal-inference-and-machine-learning-for-actionable-inference/118236/",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 do supervised machine learning and causal inference each struggle to provide actionable inference?","Question",{"text":75,"@type":76},"Supervised machine learning can be difficult to explain and may not generalize when deployed in settings different from training. Causal inference can require additional assumptions and may involve modeling high-dimensional nuisance parameters, reducing actionability.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"What main synergy does the dissertation propose between causal inference and machine learning?",{"text":80,"@type":76},"It presents methods showing how causal inference can mitigate limitations of machine learning and how machine learning can strengthen causal inference. Together, they aim to produce more trustworthy and broadly useful inference.",{"name":82,"@type":73,"acceptedAnswer":83},"What types of methods are introduced to improve causal reasoning and attribution?",{"text":84,"@type":76},"The dissertation includes inference algorithms for shift interventions, a framework for causal effects in systems with equilibrium dynamics, causal explanations for black-box behavior using causal discovery, and techniques that tighten bounds on probabilities of causation.","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"]