EEG-based Emotion Recognition through Deep Learning Models: A Comprehensive Study

Emotions play a crucial role in human communications. Emotion recognition has several applications in different fields such as brain-computer interaction, psychology, medicine, neuro-feedback, computer games, education, etc. This shows that emotion recognition is of great importance. Several modalities have been proposed for emotion recognition such as physiology-based methods, facial expression, etc. Electroencephalography (EEG)-based emotion recognition has been reported to be one of the most effective and practical emotion recognition methods. However, first studies showed lower emotion recognition accuracies. Thanks to advancement in machine learning, now it is possible to improve classification performance. Recently, numerous studies have employed advanced machine learning methods like deep learning models to enhance emotion recognition performance. In this study, several deep learning models were used to increase the performance. These methods were applied to three reliable and popular emotion datasets to draw a comprehensive comparison. We introduced four deep learning models in this paper. Results are discussed in details. Results showed that our proposed methods are effective as we achieved emotion recognition rates as high as ۹۸.۵%. We tried to conduct a comprehensive study in EEG-based emotion recognition and consider all the aspects in this paper. This paper can be used as a reference for future studies as it has employed several emotion recognition methods applied to three emotion databases.

کلیدواژه ها:

Electroencephalography ، Emotion Recognition ، EEG ، Deep Learning.