Spectral graph wavelet transform based feature representation for automated classification of emotions from EEG signal

• Published in: IEEE sensors journal Vol. 23; no. 24; p. 1
• Main Authors: Krishna, Rahul, Das, Kritiprasanna, Meena, Hemant Kumar, Pachori, Ram Bilas
• Format: Journal Article
• Language: English
• Published: New York IEEE 15.12.2023; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Accuracy; Brain modeling; Classification; Classifiers; EEG; Electrodes; Electroencephalography; Emotion recognition; Emotional factors; Emotions; Feature extraction; graph signal representation; Graphical representations; k-nearest neighbour; Machine learning; naive Bayes classifier; Principal components analysis; random forest; Sensors; Signal processing; spectral graph wavelet transform (SGWT); Wavelet transforms
• ISSN: 1530-437X; 1558-1748
• DOI: 10.1109/JSEN.2023.3330090

Electroencephalogram (EEG) monitors the brain's electrical activity and carries useful information regarding the subject's emotional states. Due to the nonstationary and being complex in nature, proper signal-processing techniques are necessary to get meaningful interpretations. The EEG signal has been represented using a graph by incorporating the temporal dependency. In this paper, a novel feature based on spectral graph wavelet transform (SGWT) for representing EEG signals has been proposed by considering the inter-dependency among different samples of EEG signals. SGWT is effective in finding multiscale information at the local level as well as global level. These multiscale representation allow to extract information about the EEG signal at different scales. The SGWT coefficients are used to develop machine-learning classifiers for emotion identification. Principle component analysis is also used for feature reduction. The proposed framework is evaluated based on a publicly available SEED dataset with the help of extensive experiments. The k-nearest neighbour classifier provides 97.3% accuracy with a standard deviation of 1.2%. The SGWT-based representation has achieved 10.0% higher accuracy compared to raw EEG signal, which shows the usefulness of the proposed approach. Our model for emotion recognition attains superior classification performance compared to state-of-the-art methods. Finally, the investigation of inter-dependency among the samples of EEG signals reveals that the SGWT-based representation of EEG signals is a useful tool for analyzing EEG signals.