EXPLORING THE POTENTIAL OF FEDERATED LEARNING TO EMPOWER CREDIT CARD FRAUDULENT TRANSACTION DETECTION WITH DEEP LEARNING TECHNIQUES

Abstract

The rapid expansion of communication systems and computing technology has led to a significant increase in both traditional and online credit card transactions. Unfortunately, this surge has also resulted in a corresponding rise in fraudulent activities, posing a serious challenge for organizations such as banking and financial institutions. To address this issue, the implementation of precise and secure transaction techniques, as well as effective fraud detection methods, becomes imperative. In this article, a novel approach utilizing a hybrid algorithmic optimization-based deep learning technique is proposed. Specifically, the Jellyfish Namib Beetle Optimization Algorithm-SpinalNet (JNBO-SpinalNet) is developed for the purpose of detecting fraudulent credit card transactions. The input data undergoes pre-processing using quantile normalization, followed by the selection of pertinent features employing diverse distance measures. To enhance the selected features, the Bootstrapping method is employed. Subsequently, the SpinalNet model is employed to identify instances of credit card fraud. The JNBO-SpinalNet model surpasses traditional detection models. The obtained results clearly demonstrate the outstanding effectiveness and efficiency of the proposed approach in identifying instances of credit card fraud.

Keywords

• federated learning
• credit card fraud detection
• optimization