Combining Mouse and Eye Movement Biometrics for User Authentication

Abstract

Previous studies show that both mouse movement and eye movement data have proven useful in authenticating a user. In this chapter, we present a user authentication system using combined features of mouse movement and eye movement. In this system, mouse movement and eye movement data are collected simultaneously and aligned based on time stamps. A set of salient features are proposed for different classification systems, including a multi-class classifier, a binary classifier, and a neural network-based regression model using fusion. Our experimental results show that the multi-class classifier works best when the number of users is small (class number = 3). For a large classification task (class number = 15), the regression model using fusion can verify a user accurately, with an average false acceptance rate (FAR) of 8.2 % and an average false rejection rate (FRR) of 6.7 %.