Evaluating Deep Learning and Machine Learning Models in Federated Learning for Credit Card Fraud Detection: A Comparative Study

This study evaluates machine learning and deep learning algorithms for credit card fraud detection within a federated learning framework. With digital banking’s rapid growth facilitating customer access yet exposing new fraud vectors, real-time detection is critical. The paper trains XGBoost and a neural network on a highly imbalanced public dataset, reflecting real fraud scenarios. Both models achieve high accuracy, but the neural network consistently outperforms XGBoost on precision, recall, and F1 score, demonstrating deep learning’s superior capability in privacy-preserving, collaborative fraud detection.