Cross-subjects Emotions Classification from EEG Signals using a Hierarchical LSTM based Classifier

This article focuses on cross subjects' emotions classification from electroencephalogram signals (EEG). We propose a hierarchical classifier based on Long Short Term Memory (LSTM) neural networks for this task. For model training and testing, we use the signals from SEED database. Cross subjects emotions classification into neutral, positive and negative achieved an accuracy of 80% when using the proposed method.