Convolutional Long Short Term Memory Model for Credit Card Detection

Abstract

At present times, credit card frauds appears as a major issue of stealing and fraud obligation using payment cards such as credit or debit cards. To prevent credit card frauds, several financial organizations were initiated to design effective models to detect the existence of frauds. The growth of Deep Learning (DL) models finds useful to examine the usual and unusual patterns along with separate transactions for raising an alert for probable frauds. This paper presents a novel DL based convolutional long short term memory (C-LSTM) model for credit fraud detection. The proposed C-LSTM model involves preprocessing, and classification. The preprocessed data undergo classification using the C-LSTM model to detect the occurrence of credit card frauds or not. The performance of the C-LSTM model is validated using German Credit and Kaggle's Credit Card Fraud Detection dataset. The attained experimental values portrayed that the C-LSTM model has shown proficient performance with the accuracy of 94% and 94.65% on the applied German credit and Credit card fraud detection dataset.