Evaluating Machine Learning and Deep Learning Models for Credit Card Fraud Detection in Cybersecurity
Md Abdul Ahad Juel, Md. Monirul Islam, Sharmin Jahan, Abu Tarek, Abdullah Al Hasib, Md. Shahadat Jaman · 2025
In this study, a robust machine learning model was developed to enhance cybersecurity in banking systems by detecting credit card fraud using the European card benchmark dataset. The research addressed the critical challenge of class imbalance through a comparative analysis, ensuring both efficiency and accuracy. Core methodologies included Multilayer Perceptron, Random Forest classifiers, with fine-tuning of hidden layers, learning rates, and epochs to optimize performance. Cross-validation techniques were applied to assess model effectiveness and mitigate overfitting. Among the tested approaches, the standard Random Forest (RF) classifier demonstrated the highest effectiveness, achieving an impressive accuracy of 99.8% and a precision of 99.2%, with no signs of overfitting. These findings underscore the potential of advanced machine learning techniques in enhancing fraud detection and strengthening cybersecurity in financial systems. The optimized RF model offers a critical solution for safeguarding sensitive transactions and combating fraudulent activities in banking.