Security Through Behavioral Biometrics and Artificial Intelligence

The purpose of this paper is to validate a method of collecting and analyzing behavioral biometric data in order to authenticate a user's identity. The method uses the time to transition from one finger to another while typing, a form of Key Interval Time biometrics (KIT). The analysis is performed using feed-forward neural nets. The User Rights and Integrity Enforcement Logic platform, URIEL, was developed to implement this methodology. The URIEL system analyzes KIT data in near-real-time. Its purpose is to demonstrate an effective means of authenticating and distinguishing amongst a group of privileged users. This concept is being assessed through live human trials. The first trial to be completed was an anonymous trial whose volunteer participants were student members of the Physics Computer Science and Engineering Department of Christopher Newport University. The high level of uniformity in this population increased the difficulty in distinguishing between individuals thereby simulating the URIEL platform's intended purpose: the regulation and identification of a small group of privileged users. The results of the first trial were encouraging and are presented here along with a discussion of the impact of variations of the methodology on the results.