Exploring personal aspects using eye-tracking modality in Tetris-playing

This paper exploits the personal aspects of an individual's eye-movements in dynamic Tetris-playing environments. Effective features representing the players' eye-moving characteristics are extracted, and they are shown to be different across difference players. Delta features are also calculated to present the dynamic changes of the static features. A series of personal identification experiments are performed by using a hidden Markov models (HMM). Our experimental results show that compared with local information, modeling and tracking the dynamic temporal information (i.e., delta features) is of more importance in distinguishing different players' eye-movement. Given a 10-zoid consecutive playing signals (about 30 seconds) we can achieve an identification rate of 82.1% by combining them both.