Modeling Temporal Behavior via Structured Hidden Markov Models: an Application to Keystroking Dynamics

Structured Hidden Markov Models (S-HMM) are a variant of Hierarchical Hidden Markov Models; it provides an abstraction mechanism allowing a high level symbolic description of the knowledge embedded in S-HMM to be easily obtained, at the same time reducing the complexity of using and learning the model. S-HMMs are particularly well suited to build up profiles of discrete processes, described by meaningful sequences of symbols possibly interleaved with gaps, i.e., subsequences whose useful information resides in their duration and not in their content. In this paper we will first introduce the model, and then we will concentrate on the description of an application, namely the characterization of biometric sequences (keyboard stroke duration) used for an identification task in computer access. Key words: Hidden Markov Model, keystroking dynamics, user authentication.

• Type: Conference
• Year: 2007
• Where: IICAI
• Authors: Ugo Galassi, Attilio Giordana, Charbel Julien, Lorenza Saitta
• Keywords: Artificial Intelligence, Hidden Markov Model, Hierarchical Hidden Markov, IICAI 2007, Level Symbolic Description