Money Transaction Fraud Detection Using Harris Grey Wolf-Based Deep Stacked Auto Encoder

Abstract

Due to the intrinsic properties of transactional data, like concept drift, noise, data imbalance, and borderline entities, the fraud detection poses a challenging issue in bank transaction. A number of solutions are developed for detecting the fraud, but these solutions reveal ineffective performance. Therefore, an effective fraud detection framework named Harris Grey Wolf (HGW)-based Deep stacked auto encoder is proposed to perform the fraud detection mechanism in bank transaction by solving the data imbalance issues. The HGW-based deep stacked auto encoder is developed using the characteristic features of the standard Harris Hawks Optimizer (HHO), and Grey Wolf Optimizer (GWO). The proposed HGW-based Deep stacked auto encoder provides an effective and optimal solution in detecting the frauds using the fitness function, which considers the minimal error value and evaluate the best solution based on the iterations. The useful and the appropriate features are effectively selected from the transactional data, as these features enhanced the accuracy of detection rate.