Fraud Detection in Real-Time Payment Systems: Ensemble Methods with Imbalanced Learning Techniques

The exponential growth of digital payment systems has led to a corresponding increase in fraudulent transactions, necessitating robust real-time fraud detection mechanisms. This paper presents a comprehensive framework combining ensemble learning methods with advanced imbalanced learning techniques to address the challenges of fraud detection in real-time payment systems. We propose a hybrid approach integrating Random Forest, Gradient Boosting, and XGBoost classifiers with Synthetic Minority Over-sampling Technique (SMOTE) and Adaptive Synthetic Sampling (ADASYN) to handle severe class imbalance. Our methodology is evaluated on two real-world datasets: the IEEE-CIS Fraud Detection dataset and the European Credit Card Transaction dataset. Experimental results demonstrate that our ensemble approach achieves 98.7% precision, 96.3% recall, and an F1-score of 97.5%, outperforming baseline methods by 12.3% in detecting fraudulent transactions while maintaining processing latency below 50ms. The proposed framework shows particular effectiveness in identifying novel fraud patterns and reducing false positives by 34.6% compared to traditional methods.

Hali tarjima qilinmagan