A New Solution to Reduce Bank Fraud Detection Fault with Particle Swarm Optimization Algorithms

Fraud and money laundering is one of the major challenges in the banking system, which makes it difficult to develop banking services. Bank fraud or money laundering-based transactions have a particular pattern and, therefore, they can be distinguished from legal transactions. However, finding these patterns is possible by data discovery methods such as data mining because most of these patterns are hidden and pattern recognition methods such as artificial neural networks are needed to detect them.Data mining techniques such as the artificial neural network are fairly good tools for detecting bank fraud, but the fault of neural network learning or other data mining methods can be significant due to the complexity of illegal transactions. Therefore, the learning of this data mining technique needs to be strengthened through methods such as metaheuristic algorithms. In this paper, particle swarm optimization was used to improve the performance of the artificial neural network and recognize the fraud pattern more accurately. The accuracy, sensitivity, and specificity in bank fraud detection were as much as 90.32%, 88.06%, and 90.12%, respectively.

Keywords:

Bank fraud , Machine learning , metaheuristic algorithm , Particle Swarm optimization algorithm