Publication

Abstract : This research investigates the use of deep learning algorithms to classify human emotions from EEG brainwave signals based on two publicly accessible Kaggle datasets. The first dataset involves EEG recordings of one male and one female subject, recorded with a Muse EEG headband on electrode locations TP9, AF7, AF8, and TP10. These signals were preprocessed with statistical feature extraction techniques. The second dataset has self-rated emotional response ratings by 14-16 participants watching 120 one-minute music video clips, rated on arousal, valence, and dominance. It further includes EEG and physiological measures from 32 participants who watched 40 individual videos and rated them. All EEG signals were processed as time-series data and marked as positive, neutral, or negative. To effectively capture both spatial and temporal dynamics of brain activity, a hybrid deep learning model combining Convolutional Neural Networks (CNN) and Gated Recurrent Units (GRU) was employed. The proposed model achieved an F1-score of 0.98 and an accuracy of 98.83%, demonstrating strong emotion recognition performance