Classification of Emotions based on EEG Signals (SEED Dataset) The basic idea of the particular implementation is to perform emotion classification from EEG signals.
As the first categorization, handcrafted features (time-domain, frequency-domain,etc.) are used, while in the second case, categorization is carried out with a combination of handcrafted and automatic features from a CNN-LSTM network.
For both cases it is used 5-fold-cross_validation.
Based on the results, we observe an increase in classification accuracy using both type of features.