[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-126824-en":3,"doc-seo-126824-105":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},126824,1099523885074,"Ivy","https://ap-avatar.wpscdn.com/davatar_9964176cb1d06d4a9deccf72a44ae3dc",8,"Research & Report","Enhancing Machine Learning Security: The Significance of Realistic Adversarial Examples - Dissertation","Adversarial attacks threaten machine learning security by using subtle, nearly invisible perturbations to mislead model decisions. Although widely investigated in vision and across domains such as credit scoring, cybersecurity, cyber-physical systems, and NLP, prior work shows limitations: many attacks produce unrealistic examples that do not match real-world objects or fail imperceptibility requirements. This thesis establishes realism as a key criterion, identifies conditions for realistic attacks, and introduces strategies that upgrade current attacks from unrealistic to realistic.","PhD-FSTM-2024-047  \nThe Faculty of Science, Technology and Medicine  \nDISSERTATION  \nDefence held on 28/06/2024 in Luxembourg  \nto obtain the degree of  \nDOCTEUR DE L’UNIVERSITÉ DU LUXEMBOURG EN  \nINFORMATIQUE  \nby  \nSalijona Dyrmishi  \nBorn on 20th December 1996 in Tërpan, Berat (Albania)  \nENHANCING MACHINE LEARNING SECURITY: THE SIGNIFICANCE OF REALISTIC ADVERSARIAL EXAMPLES  \nDissertation defence committee  \nDr. Maxime Cordy, Dissertation Supervisor Research Scientist, Université du Luxembourg  \nDr. Tegawendé Bissyandé, Chairman Associate professor, Université du Luxembourg  \nDr. Michail Papadakis, Vice Chairman Associate Professor, Université du Luxembourg  \nDr. Lorenzo Cavallaro  \nProfessor, University College London  \nDr. Yang Zhang  \nProfessor, CISPA Helmholtz Center for Information Security  \nAbstract  \nAdversarial attacks pose a significant security threat in Machine Learning (ML), employing subtle, invisible perturbations on original examples to craft instances that deceive model decisions. While extensively studied in computer vision and diverse  \n5 domains such as credit scoring, cybersecurity, cyber-physical systems, and natural language processing, recent findings reveal limitations in traditional adversarial attacks. These approaches often yield examples that lack realism, failing to map to real-world objects or adhere to imperceptibility requirements. The field of realistic adversarial attacks and their implications on the robustness of real-world systems  \n10 is currently under-explored in the literature. Through this thesis we demonstrate the importance of realism in adversarial attacks, the conditions on which these attacks are realistic, and propose new strategies to upgrade current attacks from unrealistic to realistic.  \nAs a first contribution, we shed light on the importance of producing realistic  \n15 adversarial examples when hardening models against realistic attacks. We use three real-world use cases (text classification, botnet detection, malware detection) and seven datasets in order to evaluate whether unrealistic adversarial examples can be used to protect models against realistic examples. Our results reveal discrepancies across the use cases, where unrealistic examples can either be as effective as the  \n20 realistic ones or may offer only limited improvement. Second, to explain these results, we analyze the latent representation of the adversarial examples generated with realistic and unrealistic attacks. We investigate the patterns that discriminate which unrealistic examples can be used for effective hardening.  \nAs a second contribution, we evaluate the realism of the adversarial exam-  \n25 ples generated by textual attacks. The current attacks ignore the property of imperceptibility or study it under limited settings. This entails that adversarial perturbations would not pass any human quality gate and do not represent real threats to human-checked NLP systems. To bypass this limitation and enable proper assessment (and later, improvement) of NLP model robustness, we have  \n30 surveyed 378 human participants about the perceptibility of text adversarial examples produced by state-of-the-art methods. Our results underline that existing text attacks are impractical in real-world scenarios where humans are involved.  \nThis contrasts with previous smaller-scale human studies, which reported overly optimistic conclusions regarding attacks’ success. Through our work, we hope to position human perceptibility as a first-class success criterion for text attacks, and  \nprovide guidance for research to build effective attack algorithms and, in turn, 5 design appropriate defence mechanisms.  \nAs a final contribution of this thesis, we enhance adversarial attacks based on Deep Generative Models (DGMs) by introducing a constrained layer to generate more realistic examples for tabular datasets. DGMs are widely utilized for synthesizing data that mirrors the distribution of original data, serving purpose","cbCaisxIqrSAuWKs","https://ap.wps.com/l/cbCaisxIqrSAuWKs","pdf",5142264,1,142,"English","en",105,"# Abstract\n# Adversarial Attacks and the Need for Realism\n## Evaluating Unrealistic vs Realistic Attacks\n## Latent Representation Analysis\n# Human Perceptibility in Text Adversarial Examples\n## Human Study and Practical Implications\n# Constrained Deep Generative Models for Realistic Tabular Attacks\n## Constraint Layer and Improved Attack Success","[{\"question\":\"What security risk do adversarial attacks pose in machine learning?\",\"answer\":\"They craft deceptive inputs by applying subtle, hard-to-notice perturbations that cause models to make incorrect decisions.\"},{\"question\":\"Why does the thesis focus on realistic adversarial examples?\",\"answer\":\"Traditional attacks can generate examples that lack realism or violate imperceptibility, so they may not reflect real threats in real-world settings.\"},{\"question\":\"How does the thesis improve realism in adversarial generation for different data types?\",\"answer\":\"It evaluates realism in text via human perceptibility studies, and for tabular data it enhances deep generative models by adding a constraint layer to repair violated numerical constraints.\"}]","Enhancing Machine Learning Security: The Significance of Realistic Adversarial Examples - Dissertation | PDF",1785935031,358,{"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},"enhancing-machine-learning-security-the-significance-of-realistic-adversarial-examples-dissertation","",{"@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/enhancing-machine-learning-security-the-significance-of-realistic-adversarial-examples-dissertation/126824/",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 security risk do adversarial attacks pose in machine learning?","Question",{"text":75,"@type":76},"They craft deceptive inputs by applying subtle, hard-to-notice perturbations that cause models to make incorrect decisions.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"Why does the thesis focus on realistic adversarial examples?",{"text":80,"@type":76},"Traditional attacks can generate examples that lack realism or violate imperceptibility, so they may not reflect real threats in real-world settings.",{"name":82,"@type":73,"acceptedAnswer":83},"How does the thesis improve realism in adversarial generation for different data types?",{"text":84,"@type":76},"It evaluates realism in text via human perceptibility studies, and for tabular data it enhances deep generative models by adding a constraint layer to repair violated numerical constraints.","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"]