A Systematic Review of Artificial Intelligence Methods for Fraud Detection in Electronic Payments and a Proposed Hybrid Model

The rapid growth of electronic payment systems has increased daily financial transactions and fraud complexity. AI has become a key tool for fraud detection. This paper systematically reviews methods and algorithms using AI for fraud detection in electronic payments, describing their strengths, weaknesses, and applications. National and international studies from the last five years were collected, classified, and compared regarding algorithm types, performance metrics, and implementation strategies. Supervised learning algorithms (Random Forest, XGBoost) excel at identifying fraudulent patterns, while unsupervised and anomaly detection better identify new fraud types. A hybrid approach combining both is proposed to improve accuracy and reduce false positives.

کلیدواژه ها:

Artificial Intelligence ، Machine Learning ، Fraud Detection ، Electronic Payments