Modular dynamic Bayesian network based on Markov boundary for emotion prediction in multi-sensory environment

Recently, a lot of the fields such as education, marketing, and design have applied human's emotion stimuli to increase the effectiveness of services as well as user-computer interaction. Predicting the emotion in the field is important to decide relevant stimuli because emotion has the element of uncertainty and is sensitive to sensory stimuli. In this paper, we propose a modular dynamic Bayesian network based on Markov boundary theory to predict current emotion. A relation between emotion and stimuli is identified as four types of structure. The proposed method was verified by several experiments. The computational time is 0.032 second and the average accuracy rate is 80.97%, which are quite promising for a realistic system.