According to security.org, as of 2023, 65% of credit card (CC) users in the US have been subjected to fraud at some point in their lives, which equates to about 151 million Americans. The proliferation of advanced machine learning (ML) algorithms has also contributed to detecting credit card fraud (CCF). However, using a single or static ML-based defense model against a constantly evolving adversary takes its structural advantage, which enables the adversary to reverse engineer the defense's strategy over the rounds of an iterated game. This paper proposes an adaptive moving target defense (MTD) approach based on deep reinforcement learning (DRL), termed RESONANT to identify the optimal switching points to another ML classifier for credit card fraud detection. It identifies optimal moments to strategically switch between different ML-based defense models (i.e., classifiers) to invalidate any adversarial progress and always stay a step ahead of the adversary. We take this approach in an iterated game theoretic manner where the adversary and defender take turns to take their action in the CCF detection contexts. Via extensive simulation experiments, we investigate the performance of our proposed RESONANT against that of the existing state-of-the-art counterparts in terms of the mean and variance of detection accuracy and attack success ratio to measure the defensive performance. Our results demonstrate the superiority of RESONANT over other counterparts, including static and naïve ML and MTD selecting a defense model at random (i.e., Random-MTD). Via extensive simulation experiments, our results show that our proposed RESONANT can outperform the existing counterparts up to two times better performance in detection accuracy using AUC (i.e., Area Under the Curve of the Receiver Operating Characteristic (ROC) curve) and system security against attacks using attack success ratio (ASR). / Master of Science / According to security.org, as of 2023, 65% of credit card (CC) users in the US have been subjected to fraud at some point in their lives, which equates to about 151 million Americans. The proliferation of advanced machine learning (ML) algorithms has also contributed to detecting credit card fraud (CCF). However, using a single or static ML-based defense model against a constantly evolving adversary takes its structural advantage, which enables the adversary to reverse engineer the defense's strategy over the rounds of an iterated game. This paper proposes an adaptive defense approach based on artificial intelligence (AI), termed RESONANT, to identify the optimal switching points to another ML classifiers for credit card fraud detection. It identifies optimal moments to strategically switch between different ML-based defense models (i.e., classifiers) to invalidate any adversarial progress and always stay a step ahead of the adversary. We take this approach in an iterated game theoretic manner where the adversary and defender take turns to take their action in the CCF detection contexts. Via extensive simulation experiments, we investigate the performance of our proposed RESONANT against that of the existing state-of-the-art counterparts in terms of the mean and variance of detection accuracy and attack success ratio to measure the defensive performance. Our results demonstrate the superiority of RESONANT over other counterparts, showing that our proposed RESONANT can outperform the existing counterparts by up to two times better performance in detection accuracy and system security against attacks.

近年來，信用卡市場快速地成長，發卡銀行亦大量地發卡，然而目前國內 發卡銀行在整個信用卡信用風險管理上，大都採行人類專家經驗判斷的方 式進行。發卡銀行隨著持卡人數快速地增加，其信用資料亦呈等比例急速 上升，若仍採用人工處理方式，除了會大幅增加工作負荷外，其授信品質 也不易控制。因此，本研究擬引進資訊技術來解決大量信用卡信用資料之 信用管理問題。 首先，我們探討信用卡信用管理業務，並根據其作業 流程來建構一信用卡信用管理自動化的架構，此架構包括徵信驗證系統、 審核系統、預警系統、高風險客戶管理系統、及催收系統等五個系統，其 目的在於輔助授信管理之業務、減少授管人員的工作負荷、以有效控制授 信品質、及降低授信的風險。 其次，本研究針對上述信用卡信用管理 自動化中的預警系統，利用範例學習法來建立信用卡信用風險預警範例學 習系統，且實際以一家發卡銀行的信用資料來建立並驗證四個預警模式， 期能事先讓系統自動查核信用不良之客戶。此四類預警模式為: (一)提前 預警模式(二)群體決策預警模式(三)追蹤管理預警模式(四)例外管理預警 模式 最後，我們亦提出一些未來研究之課題，期能進一步發展本研究 之信用卡信用管理自動化系統及預警模式，以推廣應用至各發卡機構。

顧客獲利性分析，係將顧客對公司之貢獻予以量化，並據以進行顧客利潤管理。藉由顧客獲利性分析，企業除了可深入瞭解獲利原因外，更可針對不同獲利性之顧客擬定不同的策略，使資源之運用更有效率，以提高整體之獲利。 □□ □□賑膍s透過與個案公司之合作，採取實地實證之研究方式，並以複迴歸模型分析銀行信用卡顧客獲利性影響因素。研究結果顯示，影響信用卡顧客獲利性之因素有往來期間、持卡等級、顧客所持正卡數、資金需求狀況、帳戶循環狀況、消費類別、信用額度使用率、顧客忠誠度…等。 □□ □□根據研究結果之隱含意涵，本研究亦對個案公司提出若干行銷與管理策略之建議，作為個案公司擬定平衡計分卡策略議題之參考。 / Customer profitability analysis is a method to manage customer profit based on the quantification of customers’ net contribution. By customer profitability analysis, companies can not only find reasons for making profit, but also draft different strategies for customers with different profitability, optimize the allocation of resources, and increase the overall profits. This study adopts field empirical method through the cooperation with the case company, and analyzes the factors influencing customer profitability by building multiple-regression models. The study results show that the factors influencing customer profitability are business period, card class, numbers of primary card, capital needs, revolving situation, categories of spending, card usage rate, customer loyalty, etc. According to the research results, this study also proposes certain suggestions about the marketing and management strategies, helping the case company frame the strategy themes in the balaced scorecard

Credit card fraud is a field with perpetrators performing illegal actions that may affect other individuals or companies negatively. For instance, a criminalcan steal credit card information from an account holder and then conduct fraudulent transactions. The activities are a potential contributory factor to how illegal organizations such as terrorists and drug traffickers support themselves financially. Within the machine learning area, there are several methods that possess the ability to detect credit card fraud transactions; supervised learning and unsupervised learning algorithms. This essay investigates the supervised approach, where two algorithms (Hellinger Distance Decision Tree (HDDT) and Random Forest) are evaluated on a real life dataset of 284,807 transactions. Under those circumstances, the main purpose is to develop a “well-functioning” model with a reasonable capacity to categorize transactions as fraudulent or legit. As the data is heavily unbalanced, reducing the false-positive rate is also an important part when conducting research in the chosen area. In conclusion, evaluated algorithms present a fairly similar outcome, where both models have the capability to distinguish the classes from each other. However, the Random Forest approach has a better performance than HDDT in all measures of interest.

Doctor of Philosophy / Department of Human Ecology-Personal Financial Planning / Kristy L. Pederson-Archuleta / The purpose of this dissertation was to examine the factors associated with students’ intentions to not possess and use a credit card. This dissertation focused on exploring a sample of undergraduate college students who do not possess a credit card. There is little known research on this group of students. The dissertation was directed by the following over-arching research question: The goal of this study was to explore college students’ beliefs about not possessing a credit card using the Theory of Reasoned Action (TRA). The research questions for this dissertation were: (a) How is personality (i.e., individual background factor) of undergraduate college students associated with their behavioral, normative, and control beliefs to not possess a credit card, (b) How are education level, age, gender, income level, religiosity, marital status, and ethnicity (i.e., social background factors) of undergraduate college students associated with their behavioral, normative, and control beliefs to not possess a credit card, and (c) How is financial knowledge (i.e., information background factor) of undergraduate college students associated with their behavioral, normative, and control beliefs to not possess a credit card. This study collected primary data. A pilot study was conducted to set the stage for the data collection of the current study. The data analysis methodology for this study consisted of the following four methods: (a) Factor Analysis, (b) Correlation Analysis, (c) MANOVA, and (d) Discriminant Function Analysis. Factor analysis identified questions were used to develop scales to measure the dependent variables. Strong reliability estimates were obtained, ranging from .84 to .94. The MANOVA test identified seven hypotheses with statistically significant results < .05. Control beliefs were significantly associated with personality. The five personality types, extraversion, agreeableness, conscientiousness, neuroticism, and openness, were all found to be significantly associated with either behavioral beliefs, control beliefs, or injunctive normative beliefs. Extraversion, agreeableness, conscientiousness, and neuroticism were all found to be associated with control beliefs. While agreeableness was also associated with injunctive normative beliefs, openness was found to be associated with behavioral beliefs. Financial knowledge was found to be associated with control beliefs. Discriminant function analysis was performed as a confirmatory test of the results from the MANOVA test, and supported the results of the MANOVA for six of the hypotheses.

Submitted by JOSIANE SANTOS DE OLIVEIRA (josianeso) on 2017-05-29T11:43:19Z No. of bitstreams: 1 Leonardo Jantsch_.pdf: 653935 bytes, checksum: 68ed1b170adfa489130eb6ea553f2c75 (MD5) / Made available in DSpace on 2017-05-29T11:43:19Z (GMT). No. of bitstreams: 1 Leonardo Jantsch_.pdf: 653935 bytes, checksum: 68ed1b170adfa489130eb6ea553f2c75 (MD5) Previous issue date: 2017-02-22 / Nenhuma / O objetivo deste trabalho foi mensurar a probabilidade de atraso nos pagamentos e posterior inadimplência como medida de análise do risco de crédito e de suporte a tomada de decisão em empréstimos de cartão de crédito para pessoas físicas em instituição financeira comercial. Como método de pesquisa buscou-se no Design Science Research a base para a prescrição de soluções e construção de artefatos, sendo que as análises foram efetivadas utilizando-se das cadeias de Markov. Os resultados encontrados evidenciam que indivíduos se comportam de forma distinta em termos de utilização e manutenção das carteiras, o que permite atribuir características próprias aos usuários de maior risco pelos atributos selecionados neste estudo. A principal contribuição deste trabalho está em evidenciar que o processo de entendimento prévio, contemplando o levantamento dos requisitos de negócio, necessidade de dados, tratamento de dados, modelagem, avaliação e implementação, pode se tornar um fator de sucesso no momento de definição e aplicação das análises de perfil por meio das cadeias de Markov. / The objective of this study was to measure the probability of late payment and subsequent delinquency as a measure of credit risk analysis and support decision making in credit card loans to individuals in a commercial financial institution. As a research method, Design Science Research was the basis for the prescription of solutions and the construction of artifacts, and the analyzes were carried out using Markov chains. The results show that individuals behave differently in terms of the use and maintenance of the portfolios, which allows to assign characteristics of the users of higher risk to the attributes selected in this study. The main contribution of this work is to show that the process of prior understanding, contemplating the survey of business requirements, data requirements, data processing, modeling, evaluation and implementation, can become a success factor when defining and applying of the profile analyzes through the Markov chains.

Made available in DSpace on 2016-04-29T14:23:30Z (GMT). No. of bitstreams: 1 Valter Lacerda de Andrade Junior.pdf: 673552 bytes, checksum: 68480511c98995570354a0166d2bb577 (MD5) Previous issue date: 2014-08-13 / The need for expert assessment of Data Mining in textual data fields and other unstructured information is increasingly present in the public and private sector. Through probabilistic models and analytical studies, it is possible to broaden the understanding of a particular information source. In recent years, technology progress caused exponential growth of the information produced and accessed in the virtual media (web and private). It is estimated that by 2003 humanity had historically generated a total of 5 exabytes of content; today that asset volume can be produced in a few days. With the increasing demand, this project aims to work with probabilistic models related to the financial market in order to check whether the textual data fields, or unstructured information, contained within the business environment, can predict certain customers behaviors. It is assumed that in the corporate environment and on the web, there is great valuable information that, due to the complexity and lack of structure, they are barely considered in probabilistic studies. This material may represent competitive and strategic advantage for business, so analyzing unstructured information one can acquire important data on behaviors and mode of user interaction in the environment in which it operates, providing data as to obtain psychographic profile and satisfaction degree. The corpus of this study consists of the results of experiments made in negotiating environment of a financial company in São Paulo. On the foregoing analysis, it was applied statistical bias semiotic concepts. Among the findings of this study, it is possible to get a critical review and thorough understanding of the processes of textual data assessment / A necessidade de análise especializada de Mineração de Dados (Data Mining) em campos textuais e em outras informações não estruturadas estão, cada vez mais, presente nas instituições dos setores públicos e privados. Por meio de modelos probabilísticos e estudos analíticos, torna-se possível ampliar o entendimento sobre determinada fonte de informação. Nos últimos anos, devido ao avanço tecnológico, observa-se um crescimento exponencial na quantidade de informação produzida e acessada nas mídias virtuais (web e privada). Até 2003, a humanidade havia gerado, historicamente, um total de 5 exabytes de conteúdo; hoje estima-se que esse volume possa ser produzido em poucos dias. Assim, a partir desta crescente demanda identificada, este projeto visa trabalhar com modelos probabilísticos relacionados ao mercado financeiro com o intuito de analisar se os campos textuais e ilustrativos, ou informações não estruturadas, contidas dentro do ambiente de negócio, podem prever certos comportamentos de clientes. Parte-se do pressuposto que, no ambiente corporativo e na web, existem informações de grande valor e que, devido à complexidade e falta de estrutura, não são consideradas em estudos probabilísticos. Isso pode representar vantagem competitiva e estratégica para o negócio, pois, por meio da análise da informação não estruturada, podem-se conhecer comportamentos e modos de interação do usuário nestes ambientes, proporcionando obter dados como perfil psicográfico e grau de satisfação. O corpus deste estudo constitui-se de resultados de experimentos efetuados no ambiente negocial de uma empresa do setor financeiro. Para as análises, foram aplicados conceitos estatísticos com viés semiótico. Entre as informações obtidas por esta pesquisa, verifica-se a compreensão crítica e aprofundada dos processos de análise textual

Nos últimos anos, tem-se observado um aumento significativo no volume de transações financeiras realizadas pela Internet. Esse crescimento no volume financeiro, associado à fragilidade inerente à ausência de verificações básicas, possíveis somente em transações do mundo físico, tem atraído a atenção de pessoas com o objetivo de obter vantagens financeiras de forma ilícita. Devido aos prejuízos causados pelas fraudes, surgiram empresas de pagamento online com o objetivo de tornar as transações de compra e venda na Internet mais seguras. Essas empresas atuam como um intermediário das transações e assumem os riscos associados, mostrando-se ser esse um negócio de alto risco. Dado o alto volume de transações com as quais essas empresas precisam lidar, torna-se clara a necessidade de métodos computacionais para detecção de transações fraudulentas, visto que a utilização estrita de verificações manuais é inviável para lidar com tal volume de transações. Essa tarefa de análise e identificação de transações fraudulentas pode ser vista como um problema computacional de classificação, sendo então aplicáveis técnicas de classificação, aprendizado computacional e mineração de dados. Porém, dada a complexidade do problema, a aplicação de técnicas computacionais só é possível após um profundo entendimento do problema e a definição de uma modelagem eficiente associada a um processo consistente e abrangente, capaz de lidar com todas as etapas necessárias para a análise eficiente de uma transação. Face a isso, o presente trabalho propõe uma abordagem abrangente para tratar o problema da fraude nesse novo mercado de intermediação de pagamentos online utilizando como base um processo já muito bem estabelecido na indústria. Abordaremos mais especificamente uma das fases desse processo, que se refere justamente a utilização de ferramentas computacionais para a detecção das fraudes, e apresentaremos um sub-processo que envolve a utilização de várias ferramentas para o tratamento do ponto de vista computacional do problema de detecção de fraudes. Para a validação dos resultados da proposta, utilizaremos uma enorme quantidade de dados reais disponibilizados por uma grande empresa do setor de intermediação de pagamentos online que colaborou com nossa pesquisa. / In recent years, there has been a significant increase in the volume of electronic transactions in the Web. This growth in trading volume, associated with the risks caused by the absence of basic checks, possible only in transactions of the physical world, has attracted the attention of people with the intention of taking advantage to obtain illicit financial benefits. Due to the injuries caused by fraud, online payment service companies emerged, with the goal of making Web transactions safer. These companies act as an intermediary between buyers and sellers, assuming all the risks, and so it is clear that it is a high-risk business. Given the high volume of transactions with which these companies must deal, it is clear the need for computational methods for detecting fraudulent transactions, as the strict use of manual checks is infeasible to handle such a volume. The task of analysis and identification of fraudulent transactions can be seen as a classification problem, and so classification, data mining and machine learning techniques can be applied to it. However, given the complexity of the problem, the application of computational techniques is only possible after a thorough understanding of the problem and the definition of an efficient model, associated with a consistent and comprehensive process which would be able to handle all the steps needed to analyze a transaction in an efficient way. Given this scenario, this work proposes a comprehensive approach to address the problem of fraud in this new business of online payment intermediation, using as basis a process already established in the industry. We will discuss more specifically one of the phases of this process, which refers to the use of computational tools to detect frauds, and we will present a sub-process using several tools to deal with the problem from a computational point of view. To validate our results, we will use a huge amount of real data provided by an important company of the online payment industry, which cooperated with our research.

由於信用卡市場開放，競爭日趨激烈，且信用卡業務與傳統一般銀行業務在行銷策略上有顯著的差異。因此，本研究以持有信用卡者為研究對象，主要目的是將持卡者作有效區隔，藉此了解不同區隔在持卡動機、使用行為變數及人口統計變數上的特徵及差異，並提出行銷策略的建議。 本研究採發放問卷方式收集資料之後，利用SAS 套裝軟體進行資料分析。首先先將持卡人所重視的信用卡特性以因素分析萃取出六個因素，接著利用此六個因素進行集群分析，將持卡者區隔成四個市場，並採用多變項變異數分析及區別分析來檢定分群的效果。然後以持有動機、使用行為變數、以及人口統計變數來描述區隔市場。 實證分析結果如下所述： 一、由持卡者所重視的信用卡特性中萃取出六個因素，分別為使用方便因素、安 全可靠因素、功能及簡便因素、信用及附卡因素、及付款及炫耀因素等。 二、由集群分析，將持卡者分為四個集群，分別為功能及簡便區隔、便利及信用區隔、安全可靠區隔、及支付區隔等。 三、不同區隔在持卡動機中的提高個人身份與地位、為了跟上流行、減少現金持有、及由於金融機構或朋友的推介等四項動機上有顯著差異。 四、不同區隔在使用行為變數中的持卡時間、使用次數、及使用場合上有顯著差異。 五、不同區隔在人口統計變數中的教育程度、職業上有顯著差異。

Submitted by Raphael Galhardo (rgalha06@gmail.com) on 2014-09-16T18:49:31Z No. of bitstreams: 1 Dissertação - RG - 2014.09.16 - 15h30.docx: 477831 bytes, checksum: 76a8788bd879c77c2bd0dbc56d418be2 (MD5) / Approved for entry into archive by Marcia Bacha (marcia.bacha@fgv.br) on 2015-02-11T18:03:35Z (GMT) No. of bitstreams: 1 Dissertação - RG - 2014.09.16 - 15h30.docx: 477831 bytes, checksum: 76a8788bd879c77c2bd0dbc56d418be2 (MD5) / Approved for entry into archive by Marcia Bacha (marcia.bacha@fgv.br) on 2015-02-11T18:05:11Z (GMT) No. of bitstreams: 1 Dissertação - RG - 2014.09.16 - 15h30.docx: 477831 bytes, checksum: 76a8788bd879c77c2bd0dbc56d418be2 (MD5) / Made available in DSpace on 2015-02-11T18:05:41Z (GMT). No. of bitstreams: 1 Dissertação - RG - 2014.09.16 - 15h30.docx: 477831 bytes, checksum: 76a8788bd879c77c2bd0dbc56d418be2 (MD5) Previous issue date: 2014-05-09 / A crise bancária que atingiu a Coreia do Sul em 2002 está relacionada, em grande parte, com a grande velocidade de crescimento e penetração da população no segmento de cartões de crédito, fazendo com que o país recebesse a denominação de 'República dos Cartões de Plástico' pela mídia internacional. Em 2003, como resultado deste crescimento, o sistema financeiro do país precisou ser socorrido pelo governo sul coreano, pois o endividamento do consumidor havia saído do controle. Atualmente, alguns estudos indicam que o Brasil pode estar caminhando para uma situação semelhante, compensando o alto nível de endividamento da população sul-coreana, com as altas taxas de juros cobradas neste segmento pelos bancos brasileiros. Com base na experiência da Coreia do Sul, este trabalho explora as principais causas deste fenômeno, a motivação da população em buscar este tipo específico de financiamento, a dicotomia dos agentes ao tratar deste crescimento e quais medidas podem ser tomadas para evitar um possível colapso no sistema financeiro. / The banking crisis that hit South Korea in 2002 is principally related to the accelerated growth and influx of credit cards used by the population, which lead to the country’s international recognition as the 'Republic of Plastic Cards' by the media. In 2003, as a result of this growth, the country's financial system had to be rescued by the South Korean government due to the consumer debt being out of control. Recent studies indicate that Brazil may be heading for a similar situation, offsetting the high level of indebtedness of the South Korean population, with high interest rates charged by Brazilian banks. Based on recent accounts in South Korea, this work explores the main causes of this phenomenon, what motivates people to seek this particular type of financing, the dichotomy of the agents to deal with this growth and what measures can be taken to prevent a possible collapse in the financial system.