Oppositional Cat Swarm Optimization-Based Feature Selection Approach for Credit Card Fraud Detection

Abstract

Credit card fraud has drastically increased in recent times due to the advancements in e-commerce systems and communication technology. Falsified credit card transactions affect the financial status of the companies as well as clients regularly and fraudsters incessantly try to develop new approaches to commit frauds. The recognition of credit card fraud is essential to sustain the trustworthiness of e-payments. Therefore, it is highly needed to design effective and accurate credit card fraud detection (CCFD) techniques. The recently developed machine learning (ML) and deep learning (DL) can be employed for CCFD because of the characteristics of building an effective model to identify fraudulent transactions. In this view, this study presents a novel oppositional cat swarm optimization-based feature selection model with a deep learning model for CCFD, called the OCSODL-CCFD technique. The major intention of the OCSODL-CCFD technique is to detect and classify fraudulent transactions using credit cards. The OCSODL-CCFD technique derives a new OCSO-based feature selection algorithm to choose an optimal subset of features. Besides, the chaotic krill herd algorithm (CKHA) with the bidirectional gated recurrent unit (BiGRU) model is applied for the classification of credit card frauds, in which the hyperparameter tuning of the BiGRU model is performed using the CKHA. To demonstrate the supreme outcomes of the OCSODL-CCFD model, a wide range of simulation analyses were carried out. The extensive comparative analysis highlighted the better outcomes of the OCSODL-CCFD model over the compared ones based on several evaluation metrics.