Abstract:
Studies on the use of alpha spectrum of an Electroencephalogram (EEG) and its relationship towards emotion, have been carried out extensively. Asymmetric Relationship Ratio (ARR) is usually used, extracted as features. To calculate the ARR, the EEG spectrum between 8 and 13 Hz needs to be averaged. Investigating the EEG alpha spectrum individually, by representing the ARR in terms of its emotional state, shows that, some EEG components produce negative relationships over others. By averaging these conflicting EEG components, it reduces the accuracy of an EEG-based emotion recognition system. This paper studies the relationship of the EEG within the alpha spectrum. Pearson, Spearman Rank and Kendall Rank correlation coefficient methods have been used to measure the dependency of the EEG components. From this study, it can be concluded that, EEG spectrums between 10 to 12 Hz are highly dependents between each other. Thus, it is proposed that only EEG within these frequency bands to be used in calculating the ARR.
