EEG Emotion Detection Using Multi-Model Classification

In the field of Human-Machine Interaction HMI identifying the human emotion is an important input for the machines to better enhance the communication. In this paper, we identifying 10 different emotions based on Valance, Arousal and Dominance using five different models and using new filters using Riemannian geometry to enhance classification. EEG signals are collected and passed to the proposed models, the accuracy of the detection was ranging from 50% to 70%. Two sessions have been conducted per subject to collect the data for training and for testing the models. The SAM assessment technique has been used to tag the training data.