Search

Copyright of Contemporary Accounting Research is the property of Canadian Academic Accounting Association and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Since the mid-1980s it has been known that Byzantine Agreement can be solved with probability 1 asynchronously, even against an omniscient, computationally unbounded adversary that can adaptively corrupt up to f < n/3 parties. Moreover, the problem is insoluble with f ≥ n/3 corruptions. However, Bracha's [13] 1984 protocol (see also Ben-Or [8]) achieved f < n/3 resilience at the cost of exponential expected latency 2Θ (n), a bound that has never been improved in this model with f = ⌊ (n-1)/3 ⌋ corruptions. In this article, we prove that Byzantine Agreement in the asynchronous, full information model can be solved with probability 1 against an adaptive adversary that can corrupt f < n/3 parties, while incurring only polynomial latency with high probability. Our protocol follows an earlier polynomial latency protocol of King and Saia [33, 34], which had suboptimal resilience, namely f ≈ n/109 [33, 34]. Resilience f = (n-1)/3 is uniquely difficult, as this is the point at which the influence of the Byzantine and honest players are of roughly equal strength. The core technical problem we solve is to design a collective coin-flipping protocol that eventually lets us flip a coin with an unambiguous outcome. In the beginning, the influence of the Byzantine players is too powerful to overcome, and they can essentially fix the coin's behavior at will. We guarantee that after just a polynomial number of executions of the coin-flipping protocol, either (a) the Byzantine players fail to fix the behavior of the coin (thereby ending the game) or (b) we can "blacklist" players such that the blacklisting rate for Byzantine players is at least as large as the blacklisting rate for good players. The blacklisting criterion is based on a simple statistical test of fraud detection. [ABSTRACT FROM AUTHOR]

In Federated Learning, Anomaly Detection poses significant challenges due to the decentralized nature of data, especially under Non-IID distributions. This study proposes a federated threshold calculation method that aggregates summary statistics from normal and anomalous data across clients to create a global threshold for Anomaly Detection with federated Autoencoders, enhancing detection accuracy and robustness while ensuring privacy. Extensive experiments on datasets, including Credit Card Fraud Detection, Shuttle, and Covertype, show that our approach consistently outperforms existing federated and local threshold calculation methods. These findings highlight the potential of summary statistics in improving federated Anomaly Detection under Non-IID conditions. [ABSTRACT FROM AUTHOR]

Dark chocolates were characterized according to geographical origin, cocoa variety, and cocoa content using the methylxanthine and polyphenolic composition and antioxidant activity as the data. The main study objective was to uncover sample patterns and identify possible markers of quality, variety, or origin to deal with authentication or fraud detection issues. In the study, a set of 26 dark chocolates from different varieties (e.g., Criollo, Forastero, and Trinitario) harvested in Africa, America, and Asia was analyzed. The optimized sample treatment consisted of defatting the chocolate (1 g of sample with 5 mL of cyclohexane for 15 min, three times) and then extracting the analytes by sonication with methanol/water 60:40 (v:v) for 15 min. The filtered extracts were analyzed by reversed‐phase high‐performance liquid chromatography with UV and spectrophotometric methods (Folin–Ciocalteu, ferric reducing antioxidant power, and aluminum methods) to determine individual phenolics and overall indexes of antioxidant and flavonoid content. Results from this chocolate set indicated that American samples are richer than African counterparts in alkaloids and phenolics (e.g., 1.7 vs. 1.1 mg g−1 caffeine and 14.5 vs. 12.5 mg g−1 total flavanols, respectively). Regarding cocoa varieties, Criollo cocoa was richer in bioactive compounds and antioxidant capacity (e.g., 16, 15, and 12 mg g−1 total flavanols for Criollo, Forastero, and Trinitario, respectively). These results indicate that the analytes resulted in potential descriptors of varietal or geographical attributes. [ABSTRACT FROM AUTHOR]

Detecting fraud accurately in credit cards is critical as this financial sector incurs significant losses for cardholders. Nonetheless, most studies adopted standard machine learning and few incremental learning, which are inadequate for addressing credit card challenges, such as rapid data arrival, unlimited data, data sensitivity, and performance decline over time. For this purpose, we propose a chunk-based incremental feature learning approach that optimises the fraud model topology for each new chunk and keeps track of one chunk each time. The model consists of several connected sub-models, where a new sub-model is optimally created for each new chunk. To avoid the network growing indefinitely, we limit the number of sub-models. To this end, we retain the most relevant sub-models to the current chunk's data distribution and re-combine them to create the optimal model. We evaluate our approach using two credit card datasets: the first of medium scale contains 2-day payments in 2013, and the second of considerable scale possesses 6-month payments in 2019. We split these datasets into multiple chunks to learn and test incrementally. We compare our approach with static learning methods trained with different scenarios. Moreover, we vary the number of historical sub-models to check their impact on the predictive performance. [ABSTRACT FROM AUTHOR]

Blockchain is an emerging technology that improves efficiency, transparency, and security in applications such as fintech, smart cities, healthcare, etc. However, blockchain technology faces scalability issues as the volume of transactions grows. One solution to enhance the scalability is offloading transactions outside the main blockchain layer using the Optimistic Rollup. In this context, we propose the Early Fraud Detection (EFD) framework that utilizes Optimistic Rollups and incorporates early fraud proofs by applying Bloom–Merkle trees that aim to reduce the challenger's verification time and cost. The EFD framework has been tested using the Ethereum Mainnet Test Network and developed with Solidity. It demonstrates that the proposed EFD framework reduces the total cost to users by 25%. Moreover, it is robust against security threats, including Double Spending, Sybil, and Denial-of-Service (DOS) attacks. [ABSTRACT FROM AUTHOR]

With the emerging trend in e-commerce, an increasing number of people have adopted cashless payment methods, especially credit cards for buying products online. However, this ever-rising usage of credit cards has also led to an increase in the malicious users attempting to gain financial profits by committing fraudulent activities resulting in huge losses to the card issuer as well as the customer. Credit Card Frauds (CCFs) are pervasive worldwide, and so efficient methods are required to detect CCFs to minimize financial losses. This research presents an efficient CCF Detection (CCFD) approach based on Deep Learning. In this work, CCFD is performed based on the features obtained from the credit card fused based on Wave Hedge distance, and the Wave Hedge coefficient utilized for fusion is estimated using the Deep Neuro-Fuzzy Network. Further, detection is performed using the Zeiler and Fergus Network (ZFNet), whose trainable factors are adjusted using the Dwarf Mongoose–Shuffled Shepherd Political Optimization (DMSSPO) algorithm. Moreover, the DMSSPO_ZFNet is analyzed based on accuracy, sensitivity, and specificity, and the experimental outcomes reveal that the values attained are 0.961, 0.961, and 0.951. [ABSTRACT FROM AUTHOR]

Although activist short sellers can play a crucial role in fraud detection, they have come under scrutiny following accusations of systematically disseminating false negative information. We develop a framework delineating the roles of campaign-specific private information quality and short-selling dynamics in shaping disclosure incentives. We predict that the act of disclosure combined with pre-disclosure stock lending dynamics is informative about the quality of an activist's private information. We find that increased pre-disclosure shorting intensity is associated with more negative post-disclosure returns, adverse media coverage, and consequential campaign outcomes, including auditor turnover, accounting restatements, class-action lawsuits, and performance-related delistings. Furthermore, elevated short-selling costs and risks magnify the association between pre-disclosure shorting intensity and post-disclosure underperformance. Finally, we examine V-shaped reversals and short covering following activists' disclosures and find no evidence of systematic manipulation. We conclude that activists disclosing fraud allegations under their own names are discouraged from engaging in "short-and-distort" schemes. Data Availability: Data are available from the sources cited in the text. JEL Classifications: G12; G14; G23; M41. [ABSTRACT FROM AUTHOR]

Financial institutions and businesses face an ongoing challenge from fraudulent transactions, prompting the need for effective detection methods. Detecting credit card fraud is crucial for identifying and preventing unauthorized transactions. While credit card fraud incidents are relatively rare, they can result in substantial financial losses, particularly due to the high monetary value associated with fraudulent transactions. Timely detection of fraud enables investigators to take swift actions to mitigate further losses. However, the investigation process is often time-consuming, limiting the number of alerts that can be thoroughly examined each day. Therefore, the primary objective of a fraud detection model is to provide accurate alerts while minimizing false alarms and missed fraud cases. In this paper, we introduce a state-of-the-art hybrid ensemble (ENS) dependable machine learning (ML) model that intelligently combines multiple algorithms with proper weighted optimization using grid search, including decision tree (DT), random forest (RF), K-nearest neighbor (KNN), and multilayer perceptron (MLP), to enhance fraud identification. To address the data imbalance issue, we employ the instant hardness threshold (IHT) technique in conjunction with logistic regression (LR), surpassing conventional approaches. Our experiments are conducted on a publicly available credit card dataset comprising 284,807 transactions. The proposed model achieves impressive accuracy rates of 99.66%, 99.73%, 98.56%, and 99.79%, and a perfect 100% for the DT, RF, KNN, MLP and ENS models, respectively. The hybrid ensemble model outperforms existing works, establishing a new benchmark for detecting fraudulent transactions in high-frequency scenarios. The results highlight the effectiveness and reliability of our approach, demonstrating superior performance metrics and showcasing its exceptional potential for real-world fraud detection applications. [ABSTRACT FROM AUTHOR]

The world is turning to financial fraud as a base for daily transactions due to the rapid growth of digital technologies, which creates numerous new opportunities for criminals to misuse credit and debit cards. The card's issuer should offer a service to shield users from any risks they might encounter in order to guarantee the security of users of those cards. This study describes the usage of deep learning algorithms for prediction of financial fraud. With the aim to represent the duration sequence generated by series of identical card transactions, we propose the use of an evolving machine learning technique. However, because most datasets containing financial fraud transactions are severely skewed, financial fraud detection remains a substantial issue for statistical solutions. This is a crucial field of research since fraudulent instances are difficult to identify and get tougher as more data is collected, decreasing the number of these instances. In this paper, utilising a Flexible Simulated Instance approach, the problem of data imbalance is resolved. Also, a tuned bidirectional LSTM with attention layer (A-BiLSTM) is proposed to detect the financial fraud. This model aims to improve existing detection strategies and detection accuracy in the context of enormous amounts of data. A benchmark dataset of financial frauds is used to evaluate the proposed model, and the outcomes are compared against existing models based on various deep learning approaches. The testing findings showed that A-BiLSTM performed flawlessly, achieving 99.96% accuracy. [ABSTRACT FROM AUTHOR]

Apart from high accuracy, what interests many researchers and practitioners in real-life tabular learning problems (e.g., fraud detection and credit scoring) is uncovering hidden patterns in the data and/or providing meaningful justification of decisions made by machine learning models. In this concern, an important question arises: should one use inherently interpretable models or explain full-complexity models such as XGBoost, Random Forest with post hoc tools? Opting for the second choice is typically supported by the accuracy metric, but it is not always evident that the performance gap is sufficiently significant, especially considering the current trend of accurate and inherently interpretable models, as well as accounting for other real-life evaluation metrics such as faithfulness, stability, and computational cost of explanations. In this work, we show through benchmarking on 45 datasets that the relative accuracy loss is less than 4% in average when using intelligible models such as explainable boosting machine. Furthermore, we propose a simple use of model ensembling to improve the expressiveness of TabSRALinear, a novel attention-based inherently interpretable solution, and demonstrate both theoretically and empirically that it is a viable option for (1) generating stable or robust explanations and (2) incorporating human knowledge during the training phase. Source code is available at https://github.com/anselmeamekoe/TabSRA. [ABSTRACT FROM AUTHOR]

For financial institutions, credit card fraud detection is a critical activity where the accuracy and efficiency of detection models are important. Traditional methods often use standard feature selection techniques that may ignore refined patterns in transaction data. This paper presents a new approach that combines feature aggregation with Exhaustive Feature Selection (EFS) to enhance the performance of credit card fraud detection models. Through feature aggregation, higher-order characteristics are created to capture complex relationships within the data, then find the most relevant features by evaluating all possible subsets of features systemically using EFS. Our method was tested using a public credit card fraud dataset, PaySim. Four popular learning classifiers—random forest (RF), decision tree (DT), logistic regression (LR), and deep neural network (DNN)—are used with balanced datasets to evaluate the techniques. The findings show a large improvement in detection accuracy, F1 score, and AUPRC compared to other approaches. Specifically, our method had improved F1 score, precision, and recall measures, which underlines its ability to handle fraudulent transactions' nuances more effectively as compared to other approaches. This article provides an overall analysis of this method's impact on model performance, giving some insights for future studies regarding fraud detection and related fields. [ABSTRACT FROM AUTHOR]

Automatic Speaker Verification (ASV) systems are crucial in various fields, enabling speaker identification for authentication, fraud detection, and forensic applications. While the simplicity and effectiveness of speech biometrics are driving the demand for ASV systems, their increasing popularity raises concerns about vulnerability to speech attacks. To enhance the security of these systems, the work in this paper proposes a spectrogram-based solution that leverages the robustness of spectrograms in audio analysis and feature extraction. The proposed model consists of two main components: frontend and backend. In the frontend, it introduces a novel spectrogram MelCochleaGram (MCG) by fusing Mel Spectrogram and Cochleagram, sequentially. For the backend implementation, pre-trained deep learning models, including ResNet50, ResNet50V2, and InceptionV3, are employed using the Keras framework. These models are individually paired with MCG to detect deepfake and replay attacks. To validate the effectiveness of the proposed system, thorough experimentation is conducted on two datasets: the DEepfake CROss-lingual (DECRO) evaluation dataset and the Voice Spoofing Detection Corpus (VSDC). The proposed combination of MCG with ResNet50 has achieved an Equal Error Rate (EER) of 0.2%, and 1.2% for deepfake detection over DECRO English and Chinese datasets, respectively. Also, for replay attack detection, the proposed combination has produced an EER of 1.4%. [ABSTRACT FROM AUTHOR]

Background and Aim: Adulteration, or the inclusion of meats not declared on the label of processed meat products, constitutes a fraudulent practice that poses a threat to public health. Sausages, which are processed meats derived from a blend of minced meats that obscure the original muscle's morphological features, are particularly prone to adulteration, making the visual detection of fraud more challenging. The research aimed to detect and measure the proportion of pork, chicken, buffalo, and beef DNA in commercially available processed meat packaged, labeled, and sold as "beef sausages" in Makassar, Indonesia. Materials and Methods: A total of 30 beef sausage samples were collected from traditional and modern markets as well as tourist attractions in Makassar. DNA was isolated and the species were identified using quantitative polymerase chain reaction. Results: The findings revealed that all sausage samples contained not only beef DNA, as indicated on their labels but also undeclared DNA from chicken and buffalo. Notably, pork DNA was not detected in the samples. The frequencies of chicken and buffalo meat were 9.2% and 10%, respectively, whereas it was 0.85% for beef in the beef sausage samples. Conclusion: The discovery of chicken and buffalo species in beef sausage samples indicates adulteration, potentially posing severe quality risks. [ABSTRACT FROM AUTHOR]

Financial fraud detection has been an urgent technical demand in cyberspace. It highly relies on clear extraction and deep representation toward complex relationships inside financial social networks. As consequence, this study combines both knowledge graph and deep learning to deal with such issue. Thus, an intelligent financial fraud detection model based on knowledge graph guidance and deep neural network is proposed in this paper. First, a new knowledge graph based on financially related systems is constructed, which includes multiple entities and relationships. Then, an adversarial learning-based neural network structure is formulated to extract financial attributes. Finally, the detection results can be output according to the extracted factors. Empirically, the proposal is implemented on a real-world dataset for performance evaluation. The experimental results show that it has higher accuracy and effectiveness compared to traditional fraud detection methods. The proposed detection model can not only identify known fraudulent behaviors, but also predict potential fraud patterns based on consumer habits, thereby improving the security and reliability of financial transactions. It can also update the knowledge graph in real-time, enabling it to cope with emerging fraud methods and variants. [ABSTRACT FROM AUTHOR]

Despite several mitigating measures, crime continues to disrupt and destabilise processes and activities in the financial sector. Detective analytics is increasingly explored as an additional mitigative solution by many financial institutions, to respond to the growing crime activities creatively and innovatively. This study aimed to examine how detective analytics can be used to follow the actors of crime activities. Actor-network theory (ANT) is employed as a lens to gain a deeper understanding of how activities can be traced, and tracked, to mitigate financial crime in institutions. The interpretive approach was applied. The findings revealed Seamless integration of incidents, Cybersecurity detection, In-house fraud detection, External infiltrate detection, and Image-matching data into one cohesive system. The study highlights the need for gaining a deeper understanding of networks of actors, following the actors, and passage points within an organisation. The findings have significant implications for improving the efficiency and effectiveness of detective analytics, from both technical and non-technical perspectives. [ABSTRACT FROM AUTHOR]

Fraud detection plays a crucial role in the modern banking sector, aiming to mitigate financial losses affecting both individuals and financial institutions. With a significant portion of the population regularly using credit cards, efforts to enhance financial inclusivity have led to increased card usage. Additionally, the rise of e-commerce has brought about a surge in credit card fraud incidents. Unfortunately, traditional statistical methods used for identifying credit card fraud are time-consuming and may not provide accurate results. As a result, machine learning algorithms have become widely adopted for effective credit card fraud detection. This study addresses the challenge of an imbalanced credit card dataset by employing three sampling strategies: cluster centroid-based majority under-sampling technique (CCMUT), synthetic minority oversampling technique (SMOTE), and oversampling technique. The training dataset is then used to train nine machine learning algorithms, including Random Forest (RF), k nearest neighbors (KNN), Decision Tree (DT), Logistic Regression (LR), Ada-boost, Extra-trees, MLP classifier, Naive Bayes, and Gradient Boosting Classifier. The performance of these approaches is assessed using metrics such as accuracy, precision, recall, f1 score, and f2 score. The dataset used in this study was obtained from the Kaggle data repository. [ABSTRACT FROM AUTHOR]

Fraud detection in banking systems is crucial for financial stability, customer protection, reputation management, and regulatory compliance. Machine Learning (ML) is vital in improving data analysis, real-time fraud detection, and developing fraud techniques by learning from data and adjusting detection strategies accordingly. Feature Selection (FS) is essential for enhancing fraud detection through ML to achieve optimal model accuracy. This is because it helps to eliminate the negative impact of redundant and irrelevant attributes. To enhance the accuracy of the given dataset, the researchers utilized multiple methods to determine the most fitting features. However, it is important to note that when implementing these methods on datasets with larger feature sizes, they may encounter issues with local optimality. Despite this, the researchers continue to work on improving the effectiveness of these methods. This study presents an effective methodology based on the Brown-Bear Optimization (BBO) algorithm to enhance the capacity to accurately identify financial CCF transactions by recognizing pertinent features. BBO has balanced capabilities to reduce dimensionality while enhancing classification accuracy. It is improved by adjusting the positions randomly to enhance exploration and exploitation capabilities, and then it is cloned into a binary variant named Binary BBOA (BBBOA). The Support Vector Machine (SVM), k-nearest Neighbor (k -NN), and Xgb-tree are the ML classifiers used with the suggested methodology. On the Australian credit dataset, the proposed methodology is compared with the basic BBOA and ten current optimizers, such as Binary African Vultures Optimization (BAVO), Binary Salp Swarm Algorithm (BSSA), Binary Atom Search Optimization (BASO), Binary Henry Gas Solubility Optimization (BHGSO), Binary Harris Hawks Optimization (BHHO), Binary Bat Algorithm (BBA), Binary Particle Swarm Optimization (BPSO), Binary Grasshopper Optimization Algorithm (BGOA), and Binary Sailfish Optimizer (BSFO). Regarding Wilcoxon's rank-sum test (α = 0. 05), the superiority and effective consequence of the presented methodology are clear on the utilized dataset and got an accuracy of classification up to 91% in the utilized dataset combined with an attribute reduction length down to 67%. The proposed methodology is further validated using 10 benchmark datasets and outperformed its competitors in most utilized datasets regarding different performance measures. In the end, the proposed methodology is further validated using ten benchmark datasets from the UCI repository. It outperformed its competitors in most of the utilized datasets regarding different performance measures. [ABSTRACT FROM AUTHOR]

The rapid advancement of Artificial Intelligence (AI) has transformed various industries, including financial auditing, by improving efficiency, accuracy, and fraud detection. This study investigates the extent of AI adoption in financial audits in Indonesia, with a focus on both Big 4 audit firms and smaller, local firms. Through a literature review and interviews with auditors from eight firms, the research explores the current state of AI utilization and the barriers to its implementation. The results indicate that while Big 4 firms are in the developmental phase of integrating AI into their auditing processes, smaller firms face significant obstacles, such as financial limitations, lack of expertise, and regulatory uncertainties, which hinder AI adoption. Despite the challenges, auditors from larger firms anticipate that AI will play a crucial role in future audits. The study concludes that AI adoption in Indonesian financial audits is uneven, and further efforts are required to support smaller firms through accessible AI tools, clearer regulations, and targeted training. These measures are essential for closing the gap in audit quality between large and small firms, ensuring broader AI implementation in the auditing sector. [ABSTRACT FROM AUTHOR]

Credit card fraud detection is a critical challenge in the financial industry, with substantial economic implications. Conventional machine learning (ML) techniques often fail to adapt to evolving fraud patterns and underperform with imbalanced datasets. This study proposes a hybrid deep learning framework that integrates Generative Adversarial Networks (GANs) with Recurrent Neural Networks (RNNs) to enhance fraud detection capabilities. The GAN component generates realistic synthetic fraudulent transactions, addressing data imbalance and enhancing the training set. The discriminator, implemented using various DL architectures, including Simple RNN, Long Short-Term Memory (LSTM) networks, and Gated Recurrent Units (GRUs), is trained to distinguish between real and synthetic transactions and further fine-tuned to classify transactions as fraudulent or legitimate. Experimental results demonstrate significant improvements over traditional methods, with the GAN-GRU model achieving a sensitivity of 0.992 and specificity of 1.000 on the European credit card dataset. This work highlights the potential of GANs combined with deep learning architectures to provide a more effective and adaptable solution for credit card fraud detection. [ABSTRACT FROM AUTHOR]

This study presents a real-time fraud detection system for online payment platforms, leveraging machine learning techniques to identify suspicious transactions. The system analyses historical transaction data to uncover patterns commonly associated with fraudulent activity. By applying algorithms such as decision trees, random forests, and logistic regression, it distinguishes between legitimate and fraudulent transactions. The system offers both user and admin interfaces: users can securely transfer funds and review their transaction history, while admins can monitor transactions and manage potential threats. Experimental results demonstrate high accuracy in fraud detection, effectively reducing false positives and issuing real-time alerts. This model, when integrated into online payment systems, enhances security and boosts user confidence in digital transactions. [ABSTRACT FROM AUTHOR]

Trust and distrust are important elements of the fiduciary relationship in the professions. Whether to trust or to distrust someone is a decision that has real consequences for success or failure of secret operations in undercover policing, in fraud investigation, or in audit. In this article, we focus on the spying profession as an extreme context in which we attempt to answer the question: how do spies navigate the trust/distrust dynamic in their work? The world of spies has often been out of bounds for those studying the professions and given that field studies in this context are extremely difficult, we analyzed biographies and autobiographies of secret agents. Based on our analysis, we identified different functions of trust and distrust: trust can be used as an instrument of manipulation and an option of last resort, while distrust is a protective mechanism aimed at shielding from vulnerability. We argue that a better understanding of trust and distrust dynamic may illuminate some of the behaviours of people in other professions. [ABSTRACT FROM AUTHOR]

Internal auditing plays a pivotal role in preventing and detecting fraudulent activities. However, the orientation and role of internal auditing in dealing with fraud risk can vary significantly across different companies. This study examines the relationship between the internal audit function (IAF) and fraud, providing new insights into the current practices of internal auditing. Using a survey dataset comprising responses from 275 Chief Audit Executives across Germany, Switzerland and Austria, we investigate factors that correlate with an increased propensity for IAFs to engage in fraud prevention and detection. Our findings suggest that a robust corporate governance environment significantly influences the extent to which the IAF is involved in preventing and detecting fraud. Shedding light on the positioning of internal auditing between management and the audit committee with respect to fraud, our results show that increased IAF involvement with management positively affects the level of activities to prevent and detect fraud, while increased IAF involvement with the audit committee has the opposite effect. Furthermore, we find that the propensity of IAFs to engage in fraud prevention and detection increases when the IAF applies technology‐based auditing techniques for risk identification. Our results have implications for building appropriate protection against the steadily increasing risk of fraud within organizations, while holistically addressing the ambiguity regarding the responsibility for preventing and detecting fraud. [ABSTRACT FROM AUTHOR]

As one of the supervised learning techniques, classification plays a crucial role in categorizing and predicting the observations across a wide range of machine learning applications such as software defect detection, fraud detection in financial sector, fault and defect detection in manufacturing industry, medical diagnosis, etc. However, most classification algorithms have been developed with the assumption that the class distribution is balanced although unequal class distributions are quite common in many practical cases. When a class imbalance problem exists, in general, the classifier tends to become biased towards the majority class and thus the minority class instances are often misclassified to the majority class. Along with the class imbalance problem, the class overlap is also known as one of the sources that makes the learning task become difficult or sometimes deteriorates the classification performance, especially, when class imbalance problem also exists. Thus, in this research, we propose a cluster impurity-based hybrid resampling method including the partially balanced strategy to improve the classification performance of class imbalanced data with considering intra-cluster class imbalance and inter-cluster overlap problems. Specifically, several clustering methods are employed for identifying the groups (i.e., clusters) of all the instances and the cluster impurity of each instance is computed for measuring the degree of cluster overlap. Then, based on the cluster impurity, the instances are generated and eliminated recursively. To demonstrate the effectiveness of the proposed method, comprehensive experiments are conducted on forty imbalanced datasets and two non-parametric hypothesis tests are employed to show the statistical difference in classification performances between the proposed method and other traditional resampling methods. [ABSTRACT FROM AUTHOR]

Credit card fraud is a widespread cybercrime that threatens financial security. Effective cybersecurity measures are essential to mitigate these risks. Machine learning has shown promising results in detecting credit card fraud by analyzing transaction data and identifying patterns of suspicious behavior. Feature selection is crucial in machine learning because it simplifies the model, improves its performance, and prevents overfitting. This research introduces a machine learning model designed for credit card fraud detection. The model makes use of three types of correlations. Pearson, Spearman, and Kendall, to identify features and enhance the fraud detection process. Testing on datasets yielded impressive results achieving category accuracies of 99.95% and 99.58% surpassing alternative approaches. Also, the results showed that Kendall correlation is the best among the three types of correlation in selecting attributes in all approved datasets. [ABSTRACT FROM AUTHOR]

Credit card fraud is a huge challenge in the financial sector, causing huge losses every year. The problem is exacerbated by increased marketing and sophisticated fraudulent activities. This study addresses the important issue of accurate real-time detection of fraudulent transactions to minimize financial losses and enhance transactional security. The main objective of this study is to develop a comprehensive fraud detection algorithm using deep learning techniques, specially designed to address the complexity and volume of modern credit card transactions. Key contributions of this research include the presentation of a new deep learning algorithm optimized for credit card fraud detection, the integration of feature engineering techniques to improve the performance of the model, and a potential scalable solution analysis in real-time Significant improvement in proven rates. The results show that the proposed deep learning-based model achieves higher accuracy and lower false positive rate, giving financial institutions a significant advantage in protecting against fraudulent activities about the character. This study highlights the power of deep learning in reforming fraud detection systems, and lays the foundation for future developments in this important area. [ABSTRACT FROM AUTHOR]

Global economies are continually threatened by financial frauds and crimes, resulting in large financial losses and ersosion of public confidence in banking institutions. Conventional methods of identifying and preventing fraud often prove ineffective due to their reactive nature and incapacity to handle large-scale transactions. This paper investigates the potential of blockchain technology as a prompt resolution for detecting and preventing financial fraud. Blockchain's inherent features of decentralization, transparency, and immutability can establish a strong framework for secure and transparent financial transactions. By utilizing the distributed ledger and smart contract features of blockchain, it is feasible to create a system that can identify fraudulent activities in real-time and prevent their occurrence. The paper explores the workings of such a system, discusses its potential advantages and challenges, and provides insights into its practical implementation. The results could herald a new age in financial security, with blockchain technology playing a crucial role in combating financial fraud and crime. In addition, the paper also delves into the integration of Hyperledger, a permissioned blockchain framework, as a strategic component in the development of the proposed system, enhancing the security and efficiency of global financial transactions. [ABSTRACT FROM AUTHOR]

Data mining techniques have proven quite effective not only in detecting financial statement frauds but also in discovering other financial crimes, such as credit card frauds, loan and security frauds, corporate frauds, bank and insurance frauds, etc. Classification of data mining techniques, in recent years, has been accepted as one of the most credible methodologies for the detection of symptoms of financial statement frauds through scanning the published financial statements of companies. The retrieved literature that has used data mining classification techniques can be broadly categorized on the basis of the type of technique applied, as statistical techniques and machine learning techniques. The biggest challenge in executing the classification process using data mining techniques lies in collecting the data sample of fraudulent companies and mapping the sample of fraudulent companies against non-fraudulent companies. In this article, a systematic literature review (SLR) of studies from the area of financial statement fraud detection has been conducted. The review has considered research articles published between 1995 and 2020. Further, a meta-analysis has been performed to establish the effect of data sample mapping of fraudulent companies against non-fraudulent companies on the classification methods through comparing the overall classification accuracy reported in the literature. The retrieved literature indicates that a fraudulent sample can either be equally paired with non-fraudulent sample (1:1 data mapping) or be unequally mapped using 1:many ratio to increase the sample size proportionally. Based on the meta-analysis of the research articles, it can be concluded that machine learning approaches, in comparison to statistical approaches, can achieve better classification accuracy, particularly when the availability of sample data is low. High classification accuracy can be obtained with even a 1:1 mapping data set using machine learning classification approaches. [ABSTRACT FROM AUTHOR]

This article examines the impact of utilizing generative artificial intelligence optimizations in automating the content generation process. This instance involves the identification of fraudulent content, which is often characterized by dynamic patterns, in addition to content production. The generated contents are constrained, which limits their dimensionality. In this scenario, duplicated contents are eliminated from the automatic creations. Furthermore, the generated ratios are utilized to discover current patterns with minimized losses and errors, hence enhancing the accuracy of generative contents. Furthermore, while analysing the created patterns, we detect a significant discrepancy in lead durations, resulting in the generation of high scores for relevant information. In order to test the results using generative tools, the adversarial network codes are employed in four scenarios. These scenarios involve generating large patterns and reducing the dynamic patterns with an enhanced accuracy of 97% in the projected model. This is in contrast to the existing approach, which only provides a content accuracy of 77% after detecting fraud. [ABSTRACT FROM AUTHOR]

This study seeks to investigate the potential of utilizing the metaverse environment to enhance the ability of financial analysts in detecting financial fraud. A preregistered randomized controlled experiment was conducted, which involved two control groups of participants discussing financial statements onsite and via video conference, and an experimental group utilizing a metaverse environment. The results revealed that the experimental group's performance in terms of the accuracy of fraud detection surpassed that of the two controlled groups. This outcome may be attributed to the enhanced usage of data visualization and more proactive participation of female participants. This study provides valuable insights into the potential benefits of employing metaverse technology in corporate finance and informs future collaboration strategies for financial analysts in fraud detection. [ABSTRACT FROM AUTHOR]

For a long time, decision-making in auditing was limited to a risk-oriented recommendation and consisted of the rigorous analysis of a sample of data. The new trend in the audit decision process focuses on the use of decision support systems (DSSs) founded on data analytics (DA) to better concentrate on in-depth analysis. This study examines how DA can improve the audit decision-making approach in the banking sector. We show that DA techniques can improve the quality of audit decision-making within banks and highlight the advantages associated with mastering these techniques, which results in a more effective and efficient audit of digital banking transactions. We propose an artifact-based data analytics-driven decision support system (DA-DSS) for an automatic fraud detection system supported by DA. The proposed DA-DSS artifact with a random forest classifier at its core is a promising innovation in the field of electronic transaction fraud detection. The results show that the random forest classifier can accurately classify the data generated by this artifact with an accuracy varying from 88 to 93% using transaction data collected from 2021 to 2022. Other classifiers including k-nearest neighbors (KNN) are also used, with a classification rate ranging from 66 to 88% for the same transaction datasets. These results show that the proposed DA-DSS with random forest can significantly improve auditing by reducing the time required for data analysis and increasing the results' accuracy. Management can use the proposed artifact to enhance and speed up the decision-making process within their organization. [ABSTRACT FROM AUTHOR]

Financial fraud negatively impacts organizational administrative processes, particularly affecting owners and/or investors seeking to maximize their profits. Addressing this issue, this study presents a literature review on financial fraud detection through machine learning techniques. The PRISMA and Kitchenham methods were applied, and 104 articles published between 2012 and 2023 were examined. These articles were selected based on predefined inclusion and exclusion criteria and were obtained from databases such as Scopus, IEEE Xplore, Taylor & Francis, SAGE, and ScienceDirect. These selected articles, along with the contributions of authors, sources, countries, trends, and datasets used in the experiments, were used to detect financial fraud and its existing types. Machine learning models and metrics were used to assess performance. The analysis indicated a trend toward using real datasets. Notably, credit card fraud detection models are the most widely used for detecting credit card loan fraud. The information obtained by different authors was acquired from the stock exchanges of China, Canada, the United States, Taiwan, and Tehran, among other countries. Furthermore, the usage of synthetic data has been low (less than 7% of the employed datasets). Among the leading contributors to the studies, China, India, Saudi Arabia, and Canada remain prominent, whereas Latin American countries have few related publications. [ABSTRACT FROM AUTHOR]

In an era where online transactions have become the norm, the battle against fraudulent activities looms large. This paper delves into the application of graph-based machine learning techniques, particularly Graph Neural Networks (GNN), in the realm of online transaction fraud detection. By leveraging the inherent network structure of transaction data, GNNs offer a powerful framework for uncovering complex patterns and anomalies indicative of fraudulent activities. In this study, we present the results obtained through the utilization of GNNs for online fraud detection, showcasing their efficacy in accurately identifying fraudulent transactions while minimizing false positives. Additionally, we discuss the implications and significance of employing graph-based machine learning techniques in enhancing fraud detection systems, emphasizing their ability to adapt to evolving fraud schemes and provide actionable insights for fraud mitigation strategies. [ABSTRACT FROM AUTHOR]

During recent decades, using credit cards represents a pivotal part of the financial lifeline. Credit cards and online payment gateways are vital elements in the world of world-wide-web. Given the fact that credit cards play an essential role in today's society, the misuse of these cards will lead to significant damages. One of the common ways to deal with these possible damages is using anomaly detection systems. These systems aim to take account of changes in customer and fraudsters' behavior to detect anomaly patterns. In the current study, we present a model namely IF-GWO to learn fraudulent patterns through analyzing past transactions. The method employs a novel ensemble learning method using isolation forest (IF) and Grey Wolf Optimizer (GWO). The experimental results indicate the priority of our presented fraud-detection system based on a noticeable number of credit card account transactions. Compared to the conventional model used for anomaly detection, the proposed model can detect more fraud accounts with fewer false positives over comparative procedures. Based on a comparison with other models using the dataset contains 284,807 transactions that are made by European cardholders, the proposed model outperformed the other approaches and achieved the highest performance in terms of F-Measure (93.52%), Area under receiver operating characteristic curve (AUC) (94.17%), and G-means (94.10%). [ABSTRACT FROM AUTHOR]

The objective of this paper is to propose a reference point in the phenomenology of poltergeists either for people who want to know more about these phenomena or for researchers looking for cases and sources associated with some particular phenomenon. In parallel, an ongoing work is conducted aimed at building a global case repository of poltergeist cases with their phenomenological characteristics and their sources, which will be available soon at www.macropk.org. A historical view of the 50+ qualitative and quantitative studies of the poltergeist phenomenon is presented along with the different authors/researchers and the publications associated. The different types of phenomena observed are studied from four angles: the physical impacts on the environment, the interactions with people, other features such as duration, focus effect, and contagion, and how the phenomena ended. Each type of event is illustrated through different cases extracted from our case repository (about 1250), often with a short extract from (one of) the sources describing some key characteristics. A discussion about the validity of these data is then developed, looking in particular at testimonials, fraud detection, legal impacts, and the similarity of description of unconnected people. These elements tend to give a strong plausibility to the diverse phenomena observed, even the more "bizarre" ones. Considering all these cases and the details associated with them could help to build a more global picture of the phenomenon. This could provide more ideas based on facts to develop current and new hypotheses, as well as new psychophysical models, in order to make progress in comprehending the phenomenon. A list of the 105 cases used in the description of the phenomenology is provided along with their sources and their distribution across historical periods and geographical areas. [ABSTRACT FROM AUTHOR]

The task of fraud detection in blockchain requires a thorough exploration of the inherent temporal and spatial characteristics in historical transaction data. Existing fraud detection methods suffer from large prediction errors. To address this issue, this paper proposed a blockchain fraud detection method, named RGCN-SRG, based on reinforced graph convolutional network (RGCN) and spatiotemporal recurrent gate (SRG). Firstly, leveraging Bitcoin's historical transaction data for the construction of the transaction graph, the method used a reinforced graph convolutional network with different kernel sizes to comprehensively extract the graph's topology information and generate feature vectors. Additionally, considering the temporal characteristics of blockchain transactions, the method introduced a spatiotemporal recurrent gate structure that incorporated graph convolutional operations into the traditional gate structure to extract dependency information from multiple spatiotemporal dimensions of the transaction graph. Finally, it obtained the prediction results of money laundering detection through a linear layer and activation function. The proposed fraud detection method was evaluated by the constructed dataset. Compared with GCN, DEDGAT, EGT and GCN + MLP F, by the proposed method improves 18.4, 10.7, 9.2 and 4.9 percentage points, respectively; the precision improves 11.5, 11.2, 7.7 and 3.7 percentage points, respectively. [ABSTRACT FROM AUTHOR]

Abnormal behavior recognition is an important technology used to detect and identify activities or events that deviate from normal behavior patterns. It has wide applications in various fields such as network security, financial fraud detection, and video surveillance. In recent years, Deep Convolution Networks (ConvNets) have been widely applied in abnormal behavior recognition algorithms and have achieved significant results. However, existing abnormal behavior detection algorithms mainly focus on improving the accuracy of the algorithms and have not explored the real-time nature of abnormal behavior recognition. This is crucial to quickly identify abnormal behavior in public places and improve urban public safety. Therefore, this paper proposes an abnormal behavior recognition algorithm based on three-dimensional (3D) dense connections. The proposed algorithm uses a multi-instance learning strategy to classify various types of abnormal behaviors, and employs dense connection modules and soft-threshold attention mechanisms to reduce the model's parameter count and enhance network computational efficiency. Finally, redundant information in the sequence is reduced by attention allocation to mitigate its negative impact on recognition results. Experimental verification shows that our method achieves a recognition accuracy of 95.61% on the UCF-crime dataset. Comparative experiments demonstrate that our model has strong performance in terms of recognition accuracy and speed. [ABSTRACT FROM AUTHOR]

The study aims to report empirical findings of quantitative research investigating what variables are significant leading proxies of fraudulent financial reporting (FFR) and the performance of the fraud detection model. The paper used financial and non-financial proxies as indicators to detect FFR with the panel data of 2235 observations of listed companies on the Vietnamese Stock Exchange from 2014 to 2020. Based on the materiality principle in auditing, the study divided the profit variance ratio into four material fraud thresholds of over 5%, 10%, 20%, and 50%. Two data mining techniques were employed: random forest for the classification model and an artificial neural network for building the best fraud prediction model. The findings show that the average accuracy of the prediction results of the random forest algorithm (RFA) reaches 91% for a material fraud threshold of 5%; when the materiality of fraud increases to above 50% of profit variance, the predictability is 98%. The average prediction accuracy of an artificial neural network (ANN) for the training set is 99%, and the test set is 97% at different fraud thresholds. These results confirm that RFA and ANN give a high accuracy in predicting fraud, and the determinants of firms committing to FFR are proxies of financial stability, followed by cash in the business and the nature of the industry. Notably, the three most important proxies related to FFR include return on total assets, return on equity, and EBT on total assets. The findings have practical implications: to identify fraudulent firms, creditors, analysts, and other stakeholders should use financial and non-financial ratios and employ data mining techniques instead of traditional fraud detection methods. [ABSTRACT FROM AUTHOR]

Many financial scandals appear to depend on a lack of skepticism on the part of their victims. Sophisticated investors trusted Bernie Madoff, for example, despite early warning signs of implausible returns. Our study investigates how education explains fraud detection in financial decision-making. In a simple survey, economics and finance students are asked to make an investment recommendation from among four hypothetical funds, including one based on Madoff's fund. We use Truth Default Theory to explain our results. We show that education increases the likelihood that students are suspicious of Madoff's fund, and that for students whose suspicions are aroused, education makes them less likely to choose the Madoff fund. [ABSTRACT FROM AUTHOR]

In the realm of credit card fraud detection, the landscape is continually evolving, demanding innovative approaches to stay ahead of increasingly sophisticated fraudulent activities. Our research pioneers a groundbreaking methodology that amalgamates the power of bipartite graph visualization with advanced machine learning techniques. This fusion yields a comprehensive framework capable of effectively evaluating the efficacy of a random forest classifier in uncovering fraudulent credit card transactions. Our study showcases the compelling application of this methodology, offering a paradigm shift in how we analyze and understand credit card fraud detection systems. By seamlessly integrating machine learning algorithms with network analysis, we provide a holistic view of the data, unveiling intricate patterns hidden within. At the heart of our approach lies the innovative use of bipartite graphs, which serve as a dynamic visual bridge between model predictions and real-world outcomes. This visual representation not only enhances interpretability but also facilitates a deeper understanding of the classifier’s performance. By visually mapping the relationships between transactions and their respective classifications, our methodology offers actionable insights into both successful detection and potential areas for improvement. Empowering analysts and stakeholders, our approach facilitates informed decisionmaking by enabling them to fine-tune model parameters and enhance the overall effectiveness of fraud detection systems. Through this synergy between cutting-edge machine learning and network analysis techniques, we provide a powerful tool to combat the critical challenge of credit card fraud prevention. Step into the future of fraud detection with our innovative methodology, where every transaction is scrutinized with precision, and where security is not just a possibility, but a promise fulfilled. [ABSTRACT FROM AUTHOR]

The most popular payment method in recent years is the credit card. Due to the E-commerce industry’s explosive growth, the usage of credit cards for online purchases have been greatly increased as a result frauds has increased. Banks have been facing challenges to detect the credit card system fraud in recent years. Credit card fraud happens when the card was stolen for any unauthorized purposes or if the fraudster utilizes the credit card information for his own use. In order to prevent credit card fraud, it is essential to build detection measures. While detecting credit card theft with machine learning (ML), the features of credit card frauds play an important and they must be carefully selected. A fraud detection algorithm must be created in order to correctly locate and stop fraudulent activity as technology advances along with the amount of fraud cases. ML methods are essential for identifying fraudulent transactions. The implementation of fraud detection models is particularly difficult because of the sensitive nature of the data, the unbalanced class distributions, and the lack of data. In this work, an intelligent approach to detect and predict online fraud transaction using extreme gradient boosting (XGBoost) algorithm is described. The XGBoost model predicts whether a transaction is fraud or not. This model will achieve better performance interarm of recall, precision, accuracy and F1-score for credit card fraud detection. [ABSTRACT FROM AUTHOR]

The sharing of data between financial institutions is widely recognised as a key component in the industry's efforts to combat fraud. Broader access to multiple sources of financial data is also critical to the development of high-quality fraud detection mechanisms based on artificial intelligence (AI). Given the challenges relating to sharing real financial data across countries and institutions, the use of synthetic data has recently become critical to enabling the exploration of broader data sharing and supporting open collaboration in AI model development. To generate synthetic data that can substitute for real data, computer algorithms closely mimic the key statistical properties of genuine data, while strictly preserving the privacy and sovereignty of the source data. This paper presents the results of an ongoing exploration into the generation of high-utility synthetic datasets of cross-border payment transactions using transformer models and discusses its application to the development of AI-based fraud prevention solutions. [ABSTRACT FROM AUTHOR]

Purpose: Although corporations exert considerable efforts to uphold ethical standards in their business operations, fraud instances persist as an enduring and formidable challenge within organisations, defying their utmost efforts. The presence of fraud poses a substantial and recurring threat to corporations, leading to significant financial losses on an annual basis. This emphasises the crucial need for a comprehensive understanding of the factors contributing to fraudulent activities and the intricate nature of fraud risk factors inherent in business operations. Therefore, this paper aims to enhance the efficacy of fraud detection and prevention measures through critical analysis and refinement of established fraud theories, drawing upon the existing literature on this subject matter. Design/methodology/approach: This paper offers a comprehensive qualitative analysis of the existing literature, thoroughly reviewing prominent models that aim to elucidate the underlying motivations behind fraudulent behaviour. Moreover, drawing upon the existing theoretical foundation, this study conceptualises a model that enhances the understanding of the crucial factors contributing to fraudulent behaviour. Findings: The study presents new theoretical insights concerning the role of personal integrity in fraudulent decision-making, presenting refined interventions that enhance comprehension of the underlying drivers of fraud occurrences and strategies for prevention. Furthermore, the study reveals a comprehensive three-part approach to improving organisational health through strengthening compliance mechanisms and cultivating an ethical-values-based culture. Originality/value: The study introduces a novel conceptual framework, the personal ethic-based fraud motivation model, which offers a deeper understanding of the factors and conditions influencing individuals' propensity to engage in fraudulent activities. Furthermore, this study presents a three Cs strategy that effectively delineates the influential forces that drive individuals to surmount fraud risks. [ABSTRACT FROM AUTHOR]

High End Money Laundering (HEML) was first introduced as a concept in the UK in 2014 and has since been held up, along with Cash Based Money Laundering (CBML), as the UK's top money laundering risk (National Crime Agency (NCA), (2014). High End Money Laundering Strategy and Action Plan, accessed via: file (nationalcrimeagency.gov.uk), on 16/07/2023; HM Government, (2023). Economic Crime Plan 2, accessed via: https://www.gov.uk/government/publications/economic-crime-plan-2023-to-2026, on 16/07/2023). Whilst most writing on CBML and HEML define these terms as relating to the nature of the original criminal proceeds (i.e., whether they were originally generated as cash or as electronic money) there is blurring and ambiguity in the terms (National Crime Agency (NCA), (2014). High End Money Laundering Strategy and Action Plan, accessed via: file (nationalcrimeagency.gov.uk), on 16/07/2023). People also talk about HEML and CBML methods, which do not necessarily relate to what form the original criminal proceeds were in (HM Government, (2023). Economic Crime Plan 2, accessed via: https://www.gov.uk/government/publications/economic-crime-plan-2023-to-2026, on 16/07/2023). This definitional confusion is underpinned by a near total absence of empirical investigations of the phenomena. In this paper I use unique access to case level detail from 31 of His Majesty's Revenue and Customs' (HMRC) Fraud Investigation Service money laundering investigations, to explore the hypothesis that high-end vs. cash-based is a meaningful disaggregation. I find that what form the original criminal proceeds were generated in does meaningfully impact patterns of subsequent money laundering and is therefore conceptually useful. I also find co-occurrence between HEML methods and CBML methods and use of cash in cases where the original proceeds were electronic. However, I argue for more precise terminology, both to enhance our understanding of these phenomena and to increase our ability to identify interdiction opportunities. This paper is an original and unpublished extension of research originally conducted by (Matanky-Becker and Cockbain, Crime, Law and Social Change 77:405–429, 2021)). [ABSTRACT FROM AUTHOR]

SUMMARY: The scope and complexity of artificial intelligence (AI) applications in auditing have grown beyond automating tasks to performing decision-making tasks. Consequently, understanding how AI-based models arrive at their decisions has become crucial, particularly for auditing tasks that demand greater accountability and that involve complex decision-making processes. In this paper, we explore the implementation of explainable AI (XAI) through a fraud detection use case and demonstrate how integrating an explainability layer using XAI can improve the interpretability of AI models, enabling stakeholders to understand the models' decision-making process. We also present emerging AI regulations in this context. [ABSTRACT FROM AUTHOR]

The financial statement auditor's identification of fraud risk factors, their assessment of fraud risk, and their fraud risk responses are key to the auditor's consideration of fraud and fraud detection. Given that the last review of research related to the search for fraud during the audit occurred nearly a decade ago, we provide a summary of recent academic research to update and inform practitioners, researchers, standard setters, regulators, and other stakeholders in the financial reporting process. We categorize and summarize findings from recent academic studies that focus on the auditor's identification, assessment, and responses to fraud risks. Implications for practice are presented for each of these areas, along with topics and questions for future research. [ABSTRACT FROM AUTHOR]

This article focuses on the problem of optimizing the system utility of Machine Learning (ML)-based systems in the presence of ML mispredictions. This is achieved via the use of self-adaptive systems and through the execution of adaptation tactics, such as model retraining, which operate at the level of individual ML components. To address this problem, we propose a probabilistic modeling framework that reasons about the cost/benefit tradeoffs associated with adapting ML components. The key idea of the proposed approach is to decouple the problems of estimating (1) the expected performance improvement after adaptation and (2) the impact of ML adaptation on overall system utility. We apply the proposed framework to engineer a self-adaptive ML-based fraud detection system, which we evaluate using a publicly available, real fraud detection dataset. We initially consider a scenario in which information on the model's quality is immediately available. Next, we relax this assumption by integrating (and extending) state-of-the-art techniques for estimating the model's quality in the proposed framework. We show that by predicting the system utility stemming from retraining an ML component, the probabilistic model checker can generate adaptation strategies that are significantly closer to the optimal, as compared against baselines such as periodic or reactive retraining. [ABSTRACT FROM AUTHOR]