[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-119456-en":3,"doc-seo-119456-105":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},119456,1374391974468,"Eden","https://ap-avatar.wpscdn.com/davatar_29158cc5080c5b710cf443261637dec0",8,"Research & Report","Adversarial Machine Learning Techniques in Fraud Detection - A Survey","Fraud Detection Systems widely deployed in companies, financial institutions, and e-commerce aim to prevent fraud and limit monetary losses. High volumes of suspected cases make manual checking impractical, so machine learning models are used, yet they remain vulnerable to adversarial attacks that exploit decision-making weaknesses. Such attacks can flip fraudulent transactions into legitimate classifications, causing direct financial harm. This thesis surveys current literature on adversarial attacks in fraud detection, maps understudied techniques and open problems, and provides a coarse-grained analysis of defensive approaches to support future research directions.","Adversarial Machine Learning Techniques in Fraud Detection: a Survey  \nTesi di Laurea Magistrale in  \nComputer Science and Engineering - Ingegneria Informatica  \nAuthor: Alessio Battaglia  \nStudent ID: 969567  \nAdvisor: Prof. Michele Carminati  \nCo-advisors: Prof. Stefano Zanero, Dr. Tommaso Paladini  \nAcademic Year: 2022-23  \ni  \nAbstract  \nFraud Detection Systems are common in companies, financial institutions, and e-commerce to prevent fraud and to reduce money loss at the minimum. The Nilson Report [2] estimated a $ 27.85 billion loss in 2018, expected to reach $40 billion by 2027 in the credit card payment sector. The FBI’s Internet Crime Complaint Center reported [1] that in 2020, they received 791,790 complaints of suspected Internet crime, with reported losses exceeding $4 .2 billion. The high number of suspected frauds makes them impossible tobe checked manually. This is why Machine Learning models are used in Fraud Detection Systems. These models are susceptible to Adversarial Attacks that exploit vulnerabilities in the model’s decision-making process. Adversarial Attacks can cause the model to misclassify a fraudulent transaction as legitimate, leading to a financial loss for the company. In this thesis, we propose a review of the current literature on adversarial attacks in the fraud detection domain. This survey aims to define what are the techniques that are not already studied (or not enough studied), giving a possible direction to future research. Moreover, we aim to highlight the problems in the research, giving a possible solution. We provided also a coarse-grained analysis of defensive techniques. This cluster is not exhaustive research but aims to give weights to adversariality effects in defensive models. Our findings about the state-of-the-art of adversarial attacks suggest a possible realization of a common framework to compare different works in the same domain. In particular, we noticed that there is no common way to evaluate attacks belonging to the same domain. Many papers are using different evaluation metrics, often without considering the success rate of an attack, a parameter that, in our opinion, is fundamental to understanding the potentiality of a threat. Moreover, we found that there are domains in which adversariality is not studied. These domains can be seen as a possible future path for the research.  \nKeywords: adversarial machine learning, survey, fraud analysis, evasion attacks, poisoning attacks  \nAbstract in lingua italiana  \nI sistemi di rilevamento delle frodi sono comuni nelle aziende, nelle istituzioni finanziarie e nell’e-commerce per prevenire le frodi e ridurre al minimo le perdite di denaro. Il Nilson Report [2] ha stimato una perdita di $ 27,85 miliardi nel 2018, che dovrebbe raggiungere $ 40 miliardi entro il 2027 nel’ambito dei pagamenti con carta di credito. L’Internet Crime Complaint Center dell’FBI ha riferito [1] che nel 2020 ha ricevuto 791.790 denunce dicrimini su Internet, con perdite dichiarate superiori a $ 4,2 miliardi. L’elevato numero di presunte frodi ne rende impossibile il controllo manuale. Questo è il motivo per cuii modelli di Machine Learning vengono utilizzati nei sistemi di rilevamento delle frodi.  \nQuesti modelli sono suscettibili agli attacchi avversariali che sfruttano le vulnerabilità nel processo decisionale del modello. Gli attacchi avversariali possono indurre il modelloa classificare erroneamente una transazione fraudolenta come legittima, portando a una perdita finanziaria per l’azienda. In questa tesi, proponiamo una revisione dell’attuale letteratura sugli attacchi adversarial nel dominio del rilevamento delle frodi. Questa indagine mira a definire quali sono le tecniche che non sono ancora studiate (o non abbastanza studiate), dando una possibile direzione alla ricerca futura. Inoltre, miriamo a evidenziare iproblemi nella ricerca, dando una possibile soluzione. Abbiamo fornito anche un’analisia grana grossa delle tecniche difensive. Questo","cbCaicWLQmjC4TKk","https://ap.wps.com/l/cbCaicWLQmjC4TKk","pdf",1667212,1,78,"English","en",105,"# 1 Introduction\n# 2 Adversarial Attacks\n## 2.1 Definition\n## 2.2 Transferability\n## 2.3 Adversarial Attacks Taxonomy\n### 2.3.1 Evasion Attacks\n### 2.3.2 Poisoning Attacks\n### 2.3.3 Exploratory Attacks\n# 3 Adversarial Attacks in Fraud Domain: State-Of-the-Art\n## 3.1 Domains of Application\n## 3.2 Datasets\n## 3.3 Adversarial Algorithms\n## 3.4 Target Models\n## 3.5 Evaluation Metrics\n## 3.6 Timeline Analysis\n## 3.7 Other Aspects\n# 4 Adversarial Defences in Fraud Domain\n## 4.1 Credit Card Domain\n### 4.1.1 Datasets","[{\"question\":\"Why are machine learning models used in fraud detection systems?\",\"answer\":\"Because the large number of suspected fraud cases makes manual verification impractical, machine learning models are employed to automate detection decisions.\"},{\"question\":\"What is the impact of adversarial attacks on fraud detection models?\",\"answer\":\"Adversarial attacks exploit weaknesses in how models make decisions, potentially causing fraudulent transactions to be misclassified as legitimate and leading to financial losses.\"},{\"question\":\"What does the thesis aim to achieve with its literature survey?\",\"answer\":\"It reviews current adversarial attack research in the fraud detection domain, highlights techniques or areas that are not sufficiently studied, and points toward directions for future work, including problems in evaluation and defenses.\"}]","Adversarial Machine Learning Techniques in Fraud Detection - A Survey | PDF",1785724369,197,{"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},"adversarial-machine-learning-techniques-in-fraud-detection-a-survey","",{"@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/adversarial-machine-learning-techniques-in-fraud-detection-a-survey/119456/",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","2026-08-03",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 are machine learning models used in fraud detection systems?","Question",{"text":75,"@type":76},"Because the large number of suspected fraud cases makes manual verification impractical, machine learning models are employed to automate detection decisions.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"What is the impact of adversarial attacks on fraud detection models?",{"text":80,"@type":76},"Adversarial attacks exploit weaknesses in how models make decisions, potentially causing fraudulent transactions to be misclassified as legitimate and leading to financial losses.",{"name":82,"@type":73,"acceptedAnswer":83},"What does the thesis aim to achieve with its literature survey?",{"text":84,"@type":76},"It reviews current adversarial attack research in the fraud detection domain, highlights techniques or areas that are not sufficiently studied, and points toward directions for future work, including problems in evaluation and defenses.","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"]