EEG Emotion Recognition Using Graph-Based CSP

Emotion recognition is a growing research field with multiple interdisciplinary applications, and processing and analyzing electroencephalogram signals (EEG) is one of its standard methods. In most articles, emotional elicitation methods for EEG signal recording involve visual-auditory stimulation; however, the use of virtual reality methods for recording signals with more realistic information is suggested. Therefore, in the present study, the VREED dataset, whose emotional elicitation is virtual reality, has been used to classify positive and negative emotions. Studies have found that the method of feature extraction significantly impacts classification accuracy. Representing EEG signals in graphs and using graph features for classification tasks has been shown to enhance accuracy by revealing hidden patterns in the signals. Additionally, the common spatial pattern (CSP) method, widely used for feature extraction in EEG signals, represents EEG signals using spatial filters and demonstrates differences between brain regions during different tasks. This paper introduces a novel graph-based CSP algorithm to incorporate the role of brain graphs into standard CSP. The graph term is added to the definition of CSP, whose graphs are learned from the relative power (RP) features. The graphs of each class positively impact calculating spatial filters that enhance the CSP method’s performance. This innovative approach achieves an $83.038 \pm 1.96$ accuracy, surpassing the previous results on the VREED dataset and presenting the algorithm’s potential for enhancing classification accuracy.