Behavioral Biometrics based Authentication System using MLP-NN and MVPA
C. V. Priya · 2021 IEEE International Power and Renewable Energy Conference (IPRECON) · 2021
In this age of information, mechanisms to ensure data and system security are of paramount importance. Password is one of the most widely used authentication mechanism, which also makes it prone to attacks. Methods to improve password authentication is a hot topic of research, in which biometric authentication offers new possibilities. Behavioral biometrics is a branch of science that studies the patterns of human activities that are uniquely recognisable and measurable. Behavioral biometrics which employs mechanisms like keystroke dynamics offers to improve password authentication without using additional hardware. This paper proposes an improvement in the existing authentication mechanism using Keystroke Dynamics (KD), Multilayer Perceptron Neural Network (MLP-NN) and Most Valuable Player Algorithm (MVPA). To overcome the drawbacks in the conventional training process, MLP-NN is trained using MVPA. Using MATLAB software, the proposed biometric authentication system is developed and validated on different users. Comparison with different authentication performance measures such FAR & FRR are also discussed.