A Robust Fraud Detection System using Deep Belief Network and Restricted Boltzmann Machines

Online security has become popular as it delivers enriched monitoring services for digital world without obstructing the usage of online services. Several techniques have been introduced to increase the quality of security systems for online faulty transactions detection. An effective model has been proposed in this research by using various domain data learning processes, deep belief networks, and restricted Boltzmann machines to provide improved transactions monitoring. Two datasets have been utilized in proposed methodology to ensure that online systems are safe and secure. The datasets used are credit card fraud detection and abstract data set for credit card fraud detection. Deep belief networks and restricted Boltzmann machines combination has helped us improve the results of proposed system. An innovative prospect to researchers has been provided by proposing a security mechanism for credit cards-based fraud detection. It has been proved that the deep belief network has been able to outperform by achieving 99.2% mean accuracy.