Credit Card Fraud Detection Using Deep Learning Approach (LSTM) Under IoT Environment

Abstract

Online payment has mainly relied since 2008 on having a rechargeable cash card that remains the most popular form of payment for e-commerce transactions. Using this card is not without risks because of its sensitivity and buying online is practical and fast yet it can be a risk when we do not know the sellers or their websites well and one of these risks is fraud, which causes great financial losses where the card holder's information is stolen and used illegally without the physical presence of the card by fraudsters who impersonate trusted institutions to steal personal or financial information via a tainted link. Most of the fraud detection systems have also shown their inability to fully detect fraudulent cards every time and repeat false alarms. In this paper one of the complex techniques known as deep learning has been used especially LSTM variant of the RNN to classify credit cards. Unlike traditional classifiers this technique is distinguished by its ability to learn. Concerning the classification by LSTM a set of information about the card is taken including time and amount.