Weightless Neural Networks for Typing Biometrics Authentication

Typing biometrics has been widely explored as a means to enhance password authentication. This paper investigates the implementation of Weightless Neural Networks (WNNs) as a pattern recognition tool to classify users’ typing patterns and thus attempt to identify the real users from impostors. In particular, we will be using a recently introduced weightless neural network, known as Deterministic RAM Network (DARN) to classify and authenticate the users based on their typing rhythms. Emphasis is also placed upon the various methods of data pre-processing to optimise the performance of the neural network for the best possible results. The experimental results cover the accuracy levels achieved through three different methods of data discretisation for comparisons.