Oral Presentation , Page 233-237 (5)

Volume Title: 2nd IEEE International Conference on Electronic Eng., Faculty of Electronic Eng., Menouf, Egypt, 3-4 July. 2021

As technology evolves rapidly, the world is using credit cards instead of cash in its everyday lives, opening up a new way for fraudulent people to abuse them. Credit card fraud losses reached approximately $28.65 billion in 2019, according to Nilsson’s report, and global card fraud is expected to reach around $32.96 billion by 2023. Providers should therefore develop an efficient model to detect and prevent fraud early. In this paper, we used deep learning techniques as an effective way to detect fraudsters in credit card transactions. Therefore, we present a model for predicting legitimate transactions or fraud on Kaggle's credit card dataset. The proposed model is OSCNN (Over Sampling with Convolution Neural Network) which is based on over-sampling preprocessing and CNN (convolution neural network). The MLP (Multi-layer perceptron) was also applied to the dataset. Comparing the MLP-OSCNN results, they proved that the proposed model achieved better results with 98% accuracy.

credit card; Deep learning; CNN; imbalanced data; and fraud detection