Emotion and Mood Recognition

Emotion and mood recognition research investigates how internal affective states can be identified from observable signals—facial muscle movements, vocal patterns, heart rate, skin conductance, and other physiological measures—often combining several of these channels at once to build a more reliable picture than any single source provides. The work sits at the intersection of cognitive psychology, which asks how emotions are structured and expressed, and machine learning, which provides tools for extracting meaningful patterns from noisy, high-dimensional data at scale. One central challenge is that emotional expression varies substantially across individuals and cultures, making it difficult to train systems that generalize beyond the datasets they were built on. Researchers are also actively working on how to handle the inherent ambiguity between short-lived emotions and slower-changing moods, and on whether models trained on acted or lab-elicited expressions transfer meaningfully to the messier, more subtle affects people display in everyday life.

Works

56,230

Total citations

533,633

Keywords

Emotion RecognitionMultimodal DataFacial ExpressionPhysiological SignalsDeep LearningAffective Computing