An Empirical Study on Interest Point Ranking and Valence-Arousal Tags of EEG Data

The research on emotional tagging has always been an important issue in the field of Affective Computing. EEG signals are ideal to the analysis of affected rate because of its strong correlation and various information for the brain activities. However, EEG signals have its own complexity and noise, the choice of electrodes and frequency band needs evidence proof from experimental analysis. This paper aims to analyze EEG signals in the frequency domain and conduct analysis to reveal the important electrodes and frequencies. Later, correlation analysis on the result and the valence-arousal tagging is conducted. The paper's result will be useful for the feature selection problem the field of Affective Computing and the automation design for valence-arousal tagging as well.