[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-124747-en":3,"doc-seo-124747-105":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},124747,16904993612988,"Olivia Brown","https://ap-avatar.wpscdn.com/davatar_a8503ba1806abce46bf441b54a3ca4cd",8,"Research & Report","The Primacy of Applied Privacy - Different Approaches to Data Privacy Needed for Different Machine Learning Tasks","This dissertation argues that data privacy requirements should be tailored to the specific machine learning task rather than handled with a single generic solution. It develops and studies multiple privacy-related approaches: examining unintended memorization in self-supervised learning, detecting data-copying in generative models, and proposing sentence-level privacy methods for document embeddings. The work provides definitions, testing or mitigation strategies, and empirical analysis to clarify when privacy risks arise and how to measure and reduce them for distinct learning settings.","UC San Diego  \nUC San Diego Electronic Theses and Dissertations  \nTitle  \nThe Primacy of Applied Privacy: Different Approaches to Data Privacy Needed for Different Machine Learning Tasks  \nPermalink  \n[https://escholarship.org/uc/item/2bn6b3h6](https://escholarship.org/uc/item/2bn6b3h6)  \nAuthor  \nMeehan, Casey  \nPublication Date  \n2023  \nPeer reviewed|Thesis/dissertation  \n[eScholarship.org](eScholarship.org) Powered by the California Digital Library  \nUniversity of California  \nUNIVERSITY OF CALIFORNIA SAN DIEGO  \nThe Primacy of Applied Privacy:  \nDifferent Approaches to Data Privacy Needed for Different Machine Learning Tasks  \nA dissertation submitted in partial satisfaction of the  \nrequirements for the degree Doctor of Philosophy  \nin  \nComputer Science  \nby  \nCasey Meehan  \nCommittee in charge:  \nProfessor Kamalika Chaudhuri, Chair  \nProfessor Taylor Berg-Kirkpatrick  \nProfessor Sanjoy Dasgupta  \nProfessor Alon Orlitsky  \nCopyright Casey Meehan, 2023 All rights reserved.  \nThe Dissertation of Casey Meehan is approved, and it is acceptable in quality and form for publication on microfilm and electronically.  \nUniversity of California San Diego  \n2023  \nDEDICATION  \nThe fact that I have made something I can write a dedication for is owed all to my parents. I cannot imagine following my heart these past few years without their unrelenting support and encouragement.  \nTABLE OF CONTENTS  \nDissertation Approval Page .................................................... iii  \nDedication .................................................................. iv  \nTable of Contents ............................................................ v  \nList of Figures ............................................................... ix  \nList of Tables ................................................................ xii  \nAcknowledgements ........................................................... xiii  \nVita ........................................................................ xv  \nAbstract of the Dissertation .................................................... xvi  \nIntroduction ................................................................. 1  \nChapter 1 Do SSL Models Have Dj Vu? A Case of Unintended Memorization in Self-supervised Learning .......................................... 3  \n1.1 Introduction ......................................................... 3  \n1.2 Preliminaries and Related Work ......................................... 5  \n1.3 Defining De´ja` Vu Memorization ........................................ 7  \n1.3.1 Testing Methodology for Measuring De´ja` Vu Memorization .......... 10  \n1.4 Quantifying De´ja` Vu Memorization ..................................... 12  \n1.4.1 Population-level Memorization ................................... 12  \n1.4.2 Sample-level Memorization ..................................... 15  \n1.5 Visualizing De´ja` Vu Memorization ...................................... 16  \n1.6 Mitigation of de´ja` vu memorization ..................................... 18  \n1.7 Conclusion .......................................................... 21  \n1.8 Acknowledgements ................................................... 21  \nChapter 2 A Non-Parametric Test to Detect Data-Copying in Generative Models .... 22  \n2.1 Introduction ......................................................... 22  \n2.1.1 Related work .................................................. 25  \n2.2 Preliminaries ........................................................ 27  \n2.2.1 Definitions of Overfitting ....................................... 27  \n2.3 A Test For Data-Copying .............................................. 29  \n2.3.1 A Global Test ................................................. 29  \n2.3.2 Handling Heterogeneity ......................................... 30  \n2.3.3 Performance Guarantees ........................................ 32  \n2.4 Experiments ......................................................... 34  \n2.4.1 Detecting da","cbCairXt8DGWXFhi","https://ap.wps.com/l/cbCairXt8DGWXFhi","pdf",32747194,1,245,"English","en",105,"# Introduction\n# Chapter 1 Do SSL Models Have Dj Vu? A Case of Unintended Memorization in Self-supervised Learning\n# Chapter 2 A Non-Parametric Test to Detect Data-Copying in Generative Models\n# Chapter 3 Sentence-level Privacy for Document Embeddings\n# Chapter 4 Privacy Implications of Shuffli","[{\"question\":\"What central claim does the dissertation make about applied privacy in machine learning?\",\"answer\":\"Privacy should be matched to the specific machine learning task, using different approaches instead of a one-size-fits-all solution.\"},{\"question\":\"How does the dissertation address unintended memorization in self-supervised learning?\",\"answer\":\"It formulates and measures déjà vu memorization, visualizes it, and studies mitigation strategies for self-supervised learning models.\"},{\"question\":\"What privacy method is proposed for document embeddings?\",\"answer\":\"The dissertation introduces sentence-level privacy for document embeddings, including definitions, embedding techniques, and a system called DeepCandidate with experiments and evaluations.\"}]","The Primacy of Applied Privacy - Different Approaches to Data Privacy Needed for Different Machine Learning Tasks | PDF",1785894268,617,{"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},"the-primacy-of-applied-privacy-different-approaches-to-data-privacy-needed-for-different-machine-learning-tasks","",{"@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/the-primacy-of-applied-privacy-different-approaches-to-data-privacy-needed-for-different-machine-learning-tasks/124747/",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","2026-08-05",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 central claim does the dissertation make about applied privacy in machine learning?","Question",{"text":75,"@type":76},"Privacy should be matched to the specific machine learning task, using different approaches instead of a one-size-fits-all solution.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"How does the dissertation address unintended memorization in self-supervised learning?",{"text":80,"@type":76},"It formulates and measures déjà vu memorization, visualizes it, and studies mitigation strategies for self-supervised learning models.",{"name":82,"@type":73,"acceptedAnswer":83},"What privacy method is proposed for document embeddings?",{"text":84,"@type":76},"The dissertation introduces sentence-level privacy for document embeddings, including definitions, embedding techniques, and a system called DeepCandidate with experiments and evaluations.","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"]