Improving Fraud Detection in Mobile Payments with Machine Learning Ensembles

Fraud detection in financial systems, especially with the rapid growth of digital and mobile payments, has become increasingly complex. Traditional rule-based systems have proven inadequate in addressing the sophisticated tactics employed by fraudsters. This study explores the application of advanced machine learning techniques to enhance fraud detection accuracy and robustness. We evaluate various models, including Logistic Regression (LR), Support Vector Machines (SVMs), XGBoost, LightGBM, and Deep Cross Network (DCN), with a focus on addressing the challenges posed by data imbalance and capturing complex feature interactions. By incorporating techniques such as Synthetic Minority Over-sampling (SMOTE) and sequence classification, the proposed approach effectively mitigates the limitations of traditional models. Additionally, interpretability tools like LIME and SHAP are employed to ensure transparency in model predictions. Experimental results demonstrate that the integration of these methods significantly improves the detection of fraudulent transactions, making the system more robust and adaptable to evolving fraud patterns. This research highlights the importance of combining diverse machine learning models and preprocessing techniques to build a comprehensive and reliable fraud detection system.