Identification of Human Emotions via Univariate and Multivarite Multiscale Entropy

This work analyzes the emotions of human in terms of complexity. This analysis is achieved by applying both univariate and multivariate multiscale entropy methods on a multimodal dataset. Most of the contemporary human-computer interaction systems are unable to identify human affective states. So, the benefit of analyzing human emotions is to fill this gap by detecting human affective states. The univariate and multivariate multiscale entropy analysis curves obtained using multimodal dataset show differences in terms of complexity among different affective states, which can be used for emotion detection and classification for machine vision applications.