Detection of credit card fraud by applying genetic algorithm and particle swarm optimization

Abstract

Fraudulent activities associated with the credit card is a pertinent problem often occurring in a global level. The customers are losing their trust with the financial institutions and the financial institutions are in a difficult state to win the goodwill of customers. A substantial number of researchers show in-terest to work on fraud detection in order to develop an optimized method or model to identify the fraudulent activities that are happening in a regular and continuous form with the credit card in our everyday life. Genetic algorithm (GA) and the potential solution-based particle swarm optimization (PSO) are two optimization algorithms, which can be considered along with the neural network to analyze the possible fraudulent transactions. The optimization algo-rithms help to make the learning process faster and optimized with a superior and better predictive accuracy value. The PSO based neural network has been trained thoroughly and performance values are compared with GA based neural network, by increasing the number of iterations and the population or number of swarms. It has been observed that algorithm based on PSO gives an optimized result for fraudulent transaction detection.