Secure Internet Financial Transactions using Multifactor Authentication and Machine Learning رسالة ماجستير

Abstract

The security of online financial transactions has emerged as a crucial concern in an era where financial services are becoming increasingly digital. The increasing use of digital platforms for banking, payments, and investment has given rise to a new wave of opportunities for both customers and cybercriminals. To address this problem, the present research offers a unique system that integrates machine learning (ML) with multifactor authentication (MFA). Using two levels of protection is the foundation of our system. The first layer uses two authentication factors, and the second layer is an embedded layer that asks for facial recognition from the user to successfully continue the purchase process if the ML model determines that the present transaction is fraudulent. To select the best classifier for constructing the ML model, four supervised ML classifiers were put into practice. After testing many classifiers, including Random Forest (RF), Decision Trees (DT), Logistic Regression (LR), and Naïve Bayes (NB), the accuracy of each was 96.717%, 97.881%, 97.938%, and 92.354%, respectively. A front-end screen for an Android e-commerce application was created to demonstrate how the framework functions. You may configure our framework to operate on any digital e-commerce platform. A thorough analysis of the body of research on the subject and various methods for securing online transactions reveals that the integration of MFA and ML has great potential for providing the greatest level of security and a system that is easy to use. In future research, it could be useful to examine other authentication factors using a different dataset.