Fraud Detection in Financial Transactions Using Deep Learning Approach: A Comparative Study

The surge in digitized financial transactions has led to an increase in financial fraud, particularly in credit card transactions, thereby propelling the need for advanced detection techniques. This research delves into the realm of fraud detection, focusing on the challenges associated with the exchange of ideas within the field. Recognizing the efficacy of machine learning in distinguishing between legitimate and fraudulent transactions, the study comprehensively reviews and evaluates existing methods employed in credit and debit fraud detection. Drawing on the efficacy of machine learning, the research evaluates prominent methods such as XGBoost, Artificial Neural Networks (ANN), and Relational Graph Convolutional Networks (R-GCN) in detecting fraudulent transactions. Drawing on insights from these method-ologies, the research identifies their strengths and limitations, paving the way for the proposition of an enhanced fraud detection technique. The findings underscore the importance of advanced analytical tools in effectively distinguishing between legitimate and deceptive patterns within financial data. This research not only contributes to the ongoing discourse on fraud detection but also provides a valuable synthesis of diverse techniques, offering a promising avenue for future advancements in the field. The three models - XGBoost, Artificial Neural Networks (ANN), and Relational Graph Convolutional Networks (R-GCN) demonstrated good results, with accuracies 97.0%, 94.0%, 98.5% respectively.