Multi-view Discriminant Analysis For Emotion Recognition

Due to the objectivity of autonomic nervous system (ANS) activity in emotional , physiological-based emotion recognition has attracted wide attention of researchers. However, emotional is multi-modal, and the use of a single modality or simply splicing multi-modal emotional data does not guarantee the accuracy of emotional recognition. Therefore, this paper proposes the use of Multi-view Discriminant Analysis Method (MDAM) for emotion recognition. Multi-modal emotional physiological data can be considered as multiple views of emotional . By maximizing the ratio of the inter-class divergence matrix to the intra-class divergence matrix, we can find multiple sets of projections, so that the projected data is lied in a discriminative common space in which the distance between the intra-class samples is minimizing, and the distance between the intra-class samples is the maximizing, which provides an effective emotional discrimination feature for multi-modal emotion recognition. The experimental results show that compared with the traditional emotion recognition method, the proposed method achieves a better recognition performance on the DEAP dataset.