AN INTELLIGENT FRAUD DETECTION SYSTEM USING GRAPH NEURAL NETWORKS AND REINFORCEMENT LEARNING

Financial fraud detection is a growing challenge in digital transactions, requiring robust solutions to identify fraudulent activities in real time. Traditional rule-based and machine learning approaches struggle to detect evolving fraud patterns, leading to high false positive rates and missed fraudulent activities. This study proposes an intelligent fraud detection system combining graph neural networks (GNNs) and reinforcement learning (RL). GNNs model transactions as a heterogeneous graph, capturing relationships between users, transactions, and financial entities. The RL component dynamically optimizes fraud detection thresholds, ensuring adaptability to new fraud tactics. Experiments on real-world financial datasets demonstrate that the proposed system outperforms traditional methods in fraud detection accuracy and adaptability. The integration of RL enables continuous learning, ensuring long-term effectiveness in combating financial fraud.