Open Access Open Access  Restricted Access Subscription or Fee Access

Credit Card Fraud Detection Using Deep Learning Techniques

R. Suvarna, A. Meena Kowshalya


Credit card frauds have become very common nowadays and lot of cases has been reported in the recent past with the increase in cybercrimes. Online credit card frauds can be efficiently detected exploiting deep-learning techniques. This paper proposes two unsupervised learning algorithms namely Auto encoder and RBM (Restricted Boltzmann Machine) that acquire information from new transactions and estimate anomalous patterns from these transactions to accurately predict number of credit card fraudulent users. Auto encoder and Restricted Boltzmann Machine were applied to credit card datasets of two continents namely Australia with 690 instances and Europe with 284,807 transactions to find the number of fraudulent users. Python libraries such as numpy, tensor flow, and sklearn were exploited to predict the accuracy, model loss and area under curve for both the datasets and the respective confusion matrix was found. The average accuracy of Auto encoder is 73% and 99% and RBM is 96% and 92% for European and Australian datasets, respectively.

Keywords: Auto encoder, confusion matrix, credit card fraud, python library, restricted boltzmann machine

Cite this Article: R Suvarna, A Meena Kowshalya. Credit card fraud detection using deep learning techniques. Journal of Web Engineering & Technology. 2020; 7 (1): 30–47.

Full Text:



  • There are currently no refbacks.