A Deep Learning Approach to Fraud Detection in Financial Payment Services

Widespread use of financial payment services besides the increased volume of financial transfers carried out in banking systems, have resulted in an important growing trend of automatic fraud detection. In this regard, an intelligent system is needed that can determine the fraudulence or genuineness of a financial transaction in real-time, unquestionably with an acceptable precision using the different transaction features. In order to gain the benefits of the system, a deep learning algorithm is described and proposed in this paper. The performance of the proposed algorithm is evaluated using a set of real-world financial transactions, which is known as the standard dataset in the literature. Then, the proposed algorithm is compared with k-nearest neighbors and support vector machine algorithms using different metrics such as accuracy, precision, F-measure, sensitivity, and precision-recall curve. The computational results confirmed the efficiency of the proposed algorithm on the standard dataset with 96% accuracy and 98% precision.