Multimodal N-Graph Feature Characteristics in Keystroke Biometrics Using Random Forest Classification

In the digital era, both general and personal data and information such as credit card numbers, banking and biometrics are starting to be digitized. This digitization carries the risk of data theft and crime, necessitating a strong and efficient security system to secure this data and information. Keystroke biometrics present an effective solution for enhancing system security, where keystrokes offer more efficient and stronger security than traditional security systems, such as passwords, one-time passwords (OTP) and single sign-on (SSO). Therefore, this study evaluates dynamic keystroke performance using the multimodal N-graph method. The research was carried out using two main scenarios, namely feature fusion classification and decision fusion classification. Based on research conducted, the use of multimodal N-graphs has better performance than using individual N-graphs in carrying out classification, with the feature fusion classification approach proving more effective than decision fusion classification.