A Deep Neural Network Algorithm for Detecting Credit Card Fraud

Abstract

The credit card is easy to use for online transaction or shopping. However, the credit card fraud happens every day all over the world and it is hard for manual detection. Many organizations including banks, information companies and even government cost billions of dollars to invent an automatic fraud detection system. In this paper, we propose a deep network for fraud detection algorithm. The log transform is used to handle the data skew problems which exists in the dataset. The focal loss is added to the network for training the hard examples. The experiments show our neural network model outperforms the other classical models like support vector machine and logistic regression.

Topics & Concepts