Artificial Intelligence in CyberSecurity: Robust RF Model Approach to Accurate Fraud Detection in Credit Card

In this study, a powerful machine learning model was developed to strengthen cyber security in banking systems by detecting fraud using the European card benchmark dataset. The study addressed the significant challenge of class imbalance through PCA-based comparative analysis, ensuring both efficiency and accuracy. Core methodologies included the use of Multilayer Perceptron, PCA-based Random Forest, and standard Random Forest classifiers, with fine-tuning of hidden layers, learning rates, and epochs to enhance performance. Cross-validation techniques were employed to evaluate model effectiveness and minimize overfitting. Among the tested approaches, the standard Random Forest (RF) classifier proved to be the most effective, achieving an impressive accuracy of 99% and a precision of 98%, with no signs of overfitting. These results highlight the potential of advanced machine learning approaches in improving fraud detection and strengthening cyber security in financial systems. The optimized RF model provides a crucial solution for protecting sensitive transactions and combating fraudulent activities in banking.