A Deep Reinforcement Learning Framework for Detecting Fraudulent Bank Account Openings

Online banking has become more popular and widespread. A huge number of customers open new bank accounts every day, which leads to a rise in fraudulent account openings. Many fraudulent activities related to this have been reported over time. It has remained a difficult task to detect fraudulent accounts efficiently. Artificial Intelligence (AI) is performing well in all the domains in the real world. So, the objective of this research is to build the most effective model to detect electronic financial transaction fraud in bank account opening. This research work aims to solve this problem (1) how to use deep reinforcement learning (DRL) for the detection of fraudulent bank accounts and (2) develop a Deep Q-Networks (DQN) based solution architecture for the detection of bank account opening fraud in an efficient way. Our proposed model achieved the highest fraud detection accuracy with 97% using the DQN algorithm with the benchmark Kaggle dataset for bank account fraud detection.