Title Of Paper:
Optimization of the Neurons and Layers Structure of the Neural Network Using the Differential Evolution Algorithm to Identify Banking System Defrauders
Author's Name :  Esmail latifi , Ehsan Gholami
KeyWords:  Fraud Detection, Difference Evolution Algorithm, Neural Network, Credit Cards
Pages:  9 -27
Volume: 5
Issue: 2
Year: 2017

The e-commerce growth has created great opportunities for world trade. It has also provided threats to banks and financial and credit institutions. One of the major challenges faced by financial institutions and banks is fraud. In financial fraud, issues such as credit card fraud, fraud in e-banking, and money laundering have attracted many concerns. The importance of using fraud detection systems has been addressed by various aspects of financial organizations. In the proposed methods, the combination of classifiers is less used, and a classification cannot provide a complete coverage of the problem space due to the imbalance of card transactions. In this research, a proposed method is applied for detecting fraud in bank credit cards through optimizing the neurons and layers structures in the neural network by using differential evolution algorithm. The proposed approach is simulated in the Matlab to nine datasets related to a Brazilian bank that includes frauds on bankcard transactions. The computational results show that the calculated cost function in the proposed method is lower than the basic method of the neural network and also has a better detection rate.

Full Text: