Multi-modal Multi-label Emotion Recognition with Heterogeneous Hierarchical Message Passing

As an important research issue in affective computing community, multi-modal emotion recognition has become a hot topic the last few years. However, almost all existing studies perform multiple binary classification for each with focus on complete time series data. In this paper, we multi-label scenario. scenario, consider not only label-to-label dependency, but also feature-to-label and modality-to-label dependencies. Particularly, propose heterogeneous hierarchical message passing network to effectively model above Furthermore, new dataset based partial time-series content show predominant generalization of our model. Detailed evaluation demonstrates effectiveness approach.