Fraud Detection in E-Transactions using Deep Neural Networks - A Case of Financial Institutions in Zimbabwe

Elliot Mbunge, Ralph Makuyana, Nation Chirara, Antony Chingosho

Abstract: Due to advancement in E-Commerce, the most common method of payment is credit card for both online and offline. It has become the most convenient way of online shopping, paying bills and money transfers. Hence, the credit card industry is investing vast amounts of money to secure credit card transactions. Financial institutions that have adopted credit card as a payment method are prone to credit card fraud attacks. The objective of this study was to develop a distributed application that analyses financial datasets to detect the possibility fraudulent activities in financial transactions. The researchers used the Hidden Markov Models (HMM) to analyze the datasets so as to generate the spending profile of a cardholder. The results generated from the HMM are then fed into the Multilayer Perceptron (MLP) that classifies the transaction into suspicious and non-suspicious classes. Since the researchers could not obtain a real dataset from the bank, one that resembles a bank dataset has been developed to train and test the MLP.

Keywords: Fraud, Credit card, E-Commerce, Deep learning, Multilayer Perceptron, Hidden Markov

How to Cite?: Elliot Mbunge, Ralph Makuyana, Nation Chirara, Antony Chingosho, "Fraud Detection in E-Transactions using Deep Neural Networks - A Case of Financial Institutions in Zimbabwe", Volume 6 Issue 9, September 2017, International Journal of Science and Research (IJSR), Pages: 1036-1041, https://www.ijsr.net/getabstract.php?paperid=ART20176804, DOI: https://dx.doi.org/10.21275/ART20176804