Active Affective State Detection and User Assistance With Dynamic Bayesian Networks.pdf

IEEE TRANSACTIONS ON SYSTEMS, MAN, AND CYBERNETICS—PART A: SYSTEMS AND HUMANS, VOL. 35, NO. 1, JANUARY 2005 93 Active Affective State Detection and User Assistance With Dynamic Bayesian Networks Xiangyang Li, Member, IEEE, and Qiang Ji, Senior Member, IEEE Abstract—With the rapid development of pervasive and ubiq- states in a timely manner, and provide assistance in terms uitous computing applications, intelligent user-assistance systems of appropriate alerts, we may prevent many accidents from face challenges of ambiguous, uncertain, and multimodal sensory happening. However, the development of such systems faces observations, user’s changing state, and various constraints on available resources and costs in making decisions. We introduce several great challenges including: 1) sensory observations of a new probabilistic framework based on the dynamic Bayesian the user are often ambiguous, uncertain, and from sources of networks (DBNs) to dynamically model and recognize user’s af- different modalities; 2) user’s affective states are often dynamic fective states and to provide the appropriate assistance in order to and evolve over time; and 3) decisions about the user’s need keep user in a productive state. We incorporate an active sensing and the assistance must be rendered appropriately and in a mechanism into the DBN framework to perform purposive and sufficing information integration in order to infer user’s affective timely and efficient manner under various constraints. state and to provide correct assistance in a timely and efficient We introduce a probabilistic framework based on the dy- manner. Experiments involving