[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-118478-en":3,"doc-seo-118478-105":30,"detail-sidebar-cat-0-en-105":92},{"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":20,"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},118478,1374391974468,"Eden","https://ap-avatar.wpscdn.com/davatar_29158cc5080c5b710cf443261637dec0",8,"Research & Report","(Un)Trustworthy Data in Adversarial Machine Learning - Dissertation","This dissertation investigates how data drives adversarial machine learning while simultaneously becoming a primary vulnerability for privacy and security. It analyzes data privacy leakage via a membership inference attack against in-context learning, showing that training usage of specific data points can be inferred even under restricted settings. It further studies data poisoning by introducing a robust poisoning technique that overcomes existing defenses, alongside the first dynamic backdoor attack using flexible triggers to evade detection. Finally, it examines how data characteristics like data importance affect attack success, yielding strategies for both attack and defense.","Saarland University  \nDepartment of Computer Science  \n(Un)Trustworthy Data in Adversarial Machine  \nLearning  \nDissertation  \nzur Erlangung des Grades  \ndes Doktors der Ingenieurwissenschaften  \nder Fakultät für Mathematik und Informatik  \nder Universität des Saarlandes  \nvon  \nRui Wen  \nSaarbrücken, 2024  \nTag des Kolloquiums: 26 März 2025  \nDekan: Prof. Dr. Roland Speicher  \nPrüfungsausschuss:  \nVorsitzender: Prof. Dr. Thorsten Herfet  \nBerichterstattende: Prof. Dr. Michael Backes  \nDr. Matthew Jagielski Prof. Dr. Tianhao Wang Dr. Yang Zhang  \nAkademischer Mitarbeiter: Dr. Mingjie Li  \nZusammenfassung  \nMaschinelles Lernen ist in verschiedenen Branchen unverzichtbar geworden, da es Innovationen vorantreibt und datengetriebene Entscheidungsprozesse ermöglicht. Im Zentrum dieser Technologie steht die zentrale Rolle von Daten, die grundlegend fürdas Modelltraining sind und die Leistung direkt beeinflussen. Allerdings macht diese Abhängigkeit von Daten maschinelle Lernsysteme auch anfällig für Schwachstellen, insbesondere im Hinblick auf Datenschutz und Sicherheit.  \nIn dieser Dissertation untersuchen wir die Rolle von Daten im adversarialen maschinellen Lernen und konzentrieren uns dabei auf zwei große Herausforderungen: Datenschutzverletzungen und Datenvergiftung. Zunächst untersuchen wir Datenschutzverletzungen in modernen Modellen, indem wir einen Membership Inference Angriff gegen In-Context Learning vorschlagen. Wir zeigen, dass es selbst in eingeschränkten Umgebungenmöglich ist, zu ermitteln, ob bestimmte Datenpunkte für das Training verwendet wurden, was erhebliche Risiken in sensiblen Bereichen wie Gesundheit und Finanzen birgt. Anschließend untersuchen wir, wie Daten als Angriffsfläche ausgenutzt werden können, indem wir eine robuste Vergiftungstechnik einführen, die derzeitige Abwehrmechanismen überwinden kann. Außerdem schlagen wir den ersten dynamischen Backdoor-Angriff vor, der flexible Trigger verwendet, um der Erkennung zu entgehen, und unterstreichendamit die Notwendigkeit stärkerer Abwehrmechanismen. Zum Schluss führen wir eine systematische Untersuchung durch, wie Datenmerkmale, wie etwa die Bedeutung von Daten, den Erfolg von Angriffen auf maschinelles Lernen beeinflussen. Unsere Ergebnissedeuten darauf hin, dass die Anpassung der Datenbedeutung entweder die Anfälligkeiterhöhen oder verringern kann, und bieten neue Strategien sowohl für Angriffe als auch für Verteidigungsmaßnahmen.  \nDiese Dissertation trägt zu einem tieferen Verständnis der adversarialen Dynamiken bei und hilft, sicherere und vertrauenswürdigere maschinelle Lernsysteme zu entwickeln.  \nAbstract  \nMachine learning has become indispensable across various industries, driving innovation and enabling data-driven decision-making. At the core of this technology is the critical role of data, which is fundamental to model training and directly impacts performance. However, this reliance on data also exposes machine learning systems to vulnerabilities, particularly around privacy and security.  \nIn this dissertation, we explore the role of data in adversarial machine learning, focusing on two major challenges: data privacy leakage and data poisoning. First, we investigate privacy leakage in state-of-the-art models by proposing a membership inference attack against in-context learning. We show that even in restricted settings, it is possible to infer whether specific data points were used in training, posing significant risks in sensitive domains such as healthcare and finance. Next, we examine how data can be exploited as an attack surface, introducing a robust poisoning technique capable of bypassing current defenses. We also propose the first dynamic backdoor attack, which uses flexible triggers to evade detection, highlighting the need for stronger defense mechanisms. Finally, we conduct a systematic study on how data characteristics, such as data importance, affect the success of machine learning attacks. Our results suggest that adjusting data i","cbCaimTWwCzeeke5","https://ap.wps.com/l/cbCaimTWwCzeeke5","pdf",7164055,1,176,"English","en",105,"# Zusammenfassung\n# Abstract\n# Background of this Dissertation\n## Published Papers and Contributions","[{\"question\":\"What are the two main challenges studied in this dissertation?\",\"answer\":\"The dissertation focuses on data privacy leakage and data poisoning in adversarial machine learning.\"},{\"question\":\"How does the proposed membership inference attack relate to in-context learning?\",\"answer\":\"It targets in-context learning and shows that, even in restricted settings, one can infer whether specific data points were used for training.\"},{\"question\":\"What makes the backdoor attack introduced in this dissertation different from prior work?\",\"answer\":\"It is a dynamic backdoor attack that uses flexible triggers to evade detection, emphasizing the need for stronger defenses.\"}]","(Un)Trustworthy Data in Adversarial Machine Learning - Dissertation | PDF",1785683802,444,{"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":87,"head_meta":89,"extra_data":91,"updated_unix":28},"untrustworthy-data-in-adversarial-machine-learning-dissertation","",{"@graph":36,"@context":86},[37,54,69],{"@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/untrustworthy-data-in-adversarial-machine-learning-dissertation/118478/",4,{"url":52,"name":13,"@type":55,"author":56,"headline":13,"publisher":58,"fileFormat":61,"inLanguage":23,"description":14,"dateModified":62,"datePublished":63,"encodingFormat":61,"isAccessibleForFree":64,"interactionStatistic":65},"DigitalDocument",{"name":9,"@type":57},"Person",{"url":41,"name":59,"@type":60},"DocShare","Organization","application/pdf","2026-08-05","2026-08-02",true,{"@type":66,"interactionType":67,"userInteractionCount":20},"InteractionCounter",{"@type":68},"ViewAction",{"@type":70,"mainEntity":71},"FAQPage",[72,78,82],{"name":73,"@type":74,"acceptedAnswer":75},"What are the two main challenges studied in this dissertation?","Question",{"text":76,"@type":77},"The dissertation focuses on data privacy leakage and data poisoning in adversarial machine learning.","Answer",{"name":79,"@type":74,"acceptedAnswer":80},"How does the proposed membership inference attack relate to in-context learning?",{"text":81,"@type":77},"It targets in-context learning and shows that, even in restricted settings, one can infer whether specific data points were used for training.",{"name":83,"@type":74,"acceptedAnswer":84},"What makes the backdoor attack introduced in this dissertation different from prior work?",{"text":85,"@type":77},"It is a dynamic backdoor attack that uses flexible triggers to evade detection, emphasizing the need for stronger defenses.","https://schema.org",{"og:url":52,"og:type":88,"og:title":13,"og:site_name":59,"og:description":14},"article",{"robots":90,"canonical":52},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":93},[94,98,102,106,111,116,121,124,129,132,136],{"id":20,"doc_module":4,"doc_module_name":46,"category_name":95,"show_sort_weight":96,"slug":97},"Story & Novel",90,"story-novel",{"id":47,"doc_module":4,"doc_module_name":46,"category_name":99,"show_sort_weight":100,"slug":101},"Literature",80,"literature",{"id":53,"doc_module":4,"doc_module_name":46,"category_name":103,"show_sort_weight":104,"slug":105},"Exam",70,"exam",{"id":107,"doc_module":4,"doc_module_name":46,"category_name":108,"show_sort_weight":109,"slug":110},5,"Comic",60,"comic",{"id":112,"doc_module":4,"doc_module_name":46,"category_name":113,"show_sort_weight":114,"slug":115},6,"Technology",50,"technology",{"id":117,"doc_module":4,"doc_module_name":46,"category_name":118,"show_sort_weight":119,"slug":120},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":122,"slug":123},30,"research-report",{"id":125,"doc_module":4,"doc_module_name":46,"category_name":126,"show_sort_weight":127,"slug":128},9,"Religion & Spirituality",20,"religion-spirituality",{"id":127,"doc_module":4,"doc_module_name":46,"category_name":130,"show_sort_weight":127,"slug":131},"World Cup","world-cup",{"id":133,"doc_module":4,"doc_module_name":46,"category_name":134,"show_sort_weight":133,"slug":135},10,"Lifestyle","lifestyle",{"id":137,"doc_module":4,"doc_module_name":46,"category_name":138,"show_sort_weight":107,"slug":139},19,"General","general"]