The AdaBoost Approach Tuned by SNS Metaheuristics for Fraud Detection

Recent advances in online payment technologies drastically increased the number of online credit card transactions, which had been additionally fueled by the recent COVID-19 pandemic. Consequently, the number of frauds related to credit cards had also increased drastically, affecting users, merchants, issuing companies, and bank institutions worldwide. One of the crucial tasks is to implement a mechanism that can assure both credit card security and integrity for each online transaction. This paper proposes an adaptive boosting algorithm, that was subjected to the optimization process by the social network search algorithm. The proposed hybrid approach has been validated on the imbalanced synthetic credit card fraud detection benchmark dataset, and acquired outcomes were compared to other cutting-edge machine learning models. The evaluation was performed by utilizing the standard performance indicators—accuracy, recall, precision, Matthews correlation coefficient, and area under the curve. The experimental findings have shown that the proposed SNS-based AdaBoost approach obtained superior results, clearly outperforming all other machine learning models included in the analysis.