Explainable Credit Card Fraud Detection with Image Conversion

The increase in the volume and velocity of credit card transactions causes class imbalance concept deviation problems data sets where fraud is detected. These make it very difficult for traditional approaches to produce robust detection models. In this study, a different perspective has been developed problem novel approach named Fraud Detection with Image Conversion (FDIC) proposed. FDIC handles as time series transforms them into images. images, which comprise temporal correlations bilateral relationships features, are classified by convolutional neural network architecture fraudulent or legitimate. When obtained results compared related studies, best F1-score recall values, 85.49% 80.35%, respectively. Since images created during process interpret, new explainable artificial intelligence also presented. way, feature that have dominant effect on revealed.