Emotion Recognition with Pre-Trained Transformers Using Multimodal Signals

Transformers employ a learned attention mechanism to dynamically score the relevance of The increasing availability of mass-market wearable devices different parts of an input according to context. Attentionbased equipped with sensors for physiological signals provides new processing is appropriate for processing physiological possibilities for monitoring the emotional health and wellbeing signals, as some parts of a signal may convey more information of users [1].