The partially observable hidden Markov model and its application to keystroke dynamics. (April 2018)

• Record Type: Journal Article
• Authors: Monaco, John V.; Tappert, Charles C.
• Is Part Of: Pattern recognition. Volume 76(2018:Apr.)
• Page Start: 449
• Page End: 462
• Publication Date: 2018-04
• Subjects: Hidden Markov model -- Keystroke biometrics -- Behavioral biometrics -- Time intervals -- Anomaly detection; Pattern perception -- Periodicals; Perception des structures -- Périodiques; Patroonherkenning; 006.4
• DOI: 10.1016/j.patcog.2017.11.021
• Languages: English
• ISSNs: 0031-3203

Abstract

Highlights: The partially observable hidden Markov model (POHMM) is introduced. In keystroke dynamics, the key names partially reveal typist behavior. The POHMM hidden state is conditioned on an independent Markov chain. The marginalized POHMM is equivalent to the HMM. A method of POHMM parameter smoothing is described. We perform user identification, verification, and continuous verification. Abstract: The partially observable hidden Markov model is an extension of the hidden Markov Model in which the hidden state is conditioned on an independent Markov chain. This structure is motivated by the presence of discrete metadata, such as an event type, that may partially reveal the hidden state but itself emanates from a separate process. Such a scenario is encountered in keystroke dynamics whereby a user's typing behavior is dependent on the text that is typed. Under the assumption that the user can be in either an active or passive state of typing, the keyboard key names are event types that partially reveal the hidden state due to the presence of relatively longer time intervals between words and sentences than between letters of a word. Using five public datasets, the proposed model is shown to consistently outperform other anomaly detectors, including the standard HMM, in biometric identification and verification tasks and is generally preferred over the HMM in a Monte Carlo goodness of fit test.