Enhancing Credit Card Fraud Detection With Sbpmc-Oa Based Feature Optimization

Credit Card Fraud Detection (CCFD) protects cardholders from money loss during transactions. However, the prevailing works could not adapt to new attack techniques, leading to fraudulent transactions. Hence, the Bias Wrapper C1 Gated Recurrent Unit (BWC1-GRU) is proposed. Initially, to train the CCFD model, the credit card transaction details are collected and pre-processed. Next, the data augmentation is done using the Synthetic Heuristic Minority Over-Sampling Technique (S-HeMOTE) technique. After augmentation, the features are extracted and using the Secretary Bird Principle of Maximum Continuous Optimization Algorithm (SBPMC-OA), optimal features are selected. Then, the CCFD is done using the BWC1-GRU method. Thus, the proposed framework detected Credit Card Fraud (CCF) with an accuracy of 97.721 %, which outperformed the existing works.