The best results were obtained by exploring a user tuned feature selection approach and a sequential classifier that uses several strokes (mouse movements between consecutive clicks) from the user. In the figure on the right we can see the effect of the interaction time on the EER. Increasing interaction times lead to decreasing error rates in user identification.
The table on the right compares some other behavioral biometric techniques with the proposed mouse movement biometric with different strokes length durations. Given the performance of the proposed biometric technique, there are some situations, like continuous biometric applications, that the performance is comparable to the other popular behavioural biometrics techniques, with the advantages of low intrusion, no need for special sensors, and the capability of being remotelly colected. We also consider this trait usefull for integration on multibiometrics systems.
Hugo Gamboa and Vasco Ferreira, Widam - web interaction display and monitoring. 5th International Conference on Enterprise Information Systems, ICEIS 2003, Angers, France, 2003.
Hugo Gamboa and Ana Fred, A behavioral biometric system based on human-computer interaction, Proceedings of SPIE Vol. 5404, p. 381-392, Biometric Technology for Human Identification; Anil K. Jain, Nalini K. Ratha; Eds. Orlando USA, 2004.
Hugo Gamboa, Ana Fred and António Alves Vieira, Prevention or Identification of Web Intrusion via Human Computer Interaction Behaviour - A Proposal, Meeting Prooceedings of Nato Research and Technology Organization Symposium on Systems, Concepts and Integration (SCI) Methods and Technologies for Defence Against Terrorism, RTO-MP-SCI-158, London, United Kingdom, 2004.
Hugo Gamboa, Ana Fred and Anil K.Jain, Webbiometrics: User Verification Via Web Interaction,. Biometric Symposium, BCC, Baltimore, USA, 2007 (Finalist of the EBF European Biometrics Research Award 2007, the work having been presented on October 3, Brussels, Belgium).