Machine Learning
Abstract
Emotion recognition is an important research topic.Physiological signals seem to be an appropriate way for emotion recognition and specific sensors are required to collect these data.Therefore, laboratory sensors are commonly used while the number of wearable devices including similar physiological sensors is growing up.Many studies have been completed to evaluate the signal quality obtained by these sensors but without focusing on their emotion recognition capabilities.In the current study, Machine Learning models were trained to compare the Biopac MP150 (laboratory sensor) and Empatica E4 (wearable sensor) in terms of emotion recognition accuracy.Results show similar accuracy between data collected using laboratory and wearable sensors.These results support the reliability of emotion recognition outside laboratory.