Machine learning-based hybrid emotions recognition model using electroencephalogram signals

This paper uses Hindi video clips to propose an electroencephalogram (EEG) signal-based hybrid system for emotion identification. EEG signals cannot be altered, unlike other forms of expressiveness-like voice and facial emotion. The suggested approach uses a self-created dataset under the control environments. Accuracy is the main objective of the proposed model. This study used a self-created constructed using an 8-channel unicorn black hybrid EEG machine on 30 participants while they viewed Hindi movie video clips mimicking emotions: happy, fearful, sad, and neutral. The proposed model used a two-hybrid classifier support vector machine (SVM) and k-nearest neighbor (KNN), implemented using MATLAB R2017a. In the proposed implementation, the four emotion classification categories (happy, sad, fear, and neutral) observed an average accuracy of 60.832%. The results of the presented study were compared with two recent systems. It was found that the proposed system observed better accuracy for the category of NHP five classes and the category of HP Five Classes.