Learning Affective Features With a Hybrid Deep Model for Audio–Visual Emotion Recognition

Emotion recognition is challenging due to the emotional gap between emotions and audio-visual features. Motivated by the powerful feature learning ability of deep neural networks, this paper proposes to bridge the emotional gap by using a hybrid deep model, which first produces audio-visual segment ... View More