A practical approach to user authentication

A method for user authentication is presented which analyzes keystroking data as the user types his or her name. This study utilizes the ADALINE (ADAptive LINear Element) and backpropagation neural nets to identify the typing pattern characteristic of a particular user. A simple measure of geometric distance is also used for comparison. This paper provides a brief introduction to this type of neural net. It then describes the research procedure and contrasts the initial and new results, followed by a conclusion with notes concerning future research. For an average 15-character name, a complete exclusion of imposters is obtained from a set of over 5000 imposter samples.<
>