Subject independent emotion recognition using EEG and physiological signals – a comparative study

Purpose The aim of this study is to investigate subject independent emotion recognition capabilities EEG and peripheral physiological signals namely: electroocoulogram (EOG), electromyography (EMG), electrodermal activity (EDA), temperature, plethysmograph respiration. experiments are conducted on both modalities independently in combination. This arranges the order based prediction accuracy obtained test data using time frequency domain features. Design/methodology/approach DEAP dataset used experiment. Time features extracted, followed by correlation-based feature selection. Classifiers namely – Naïve Bayes, logistic regression, linear discriminant analysis, quadratic logit boost stacking trained selected Based performance classifiers set, best modality for each dimension identified. Findings experimental results with as one all another indicate that better at arousal compared 7.18%, while valence 3.51%. EOG superior zygomaticus (zEMG) EDA 1.75% cost higher number electrodes. paper concludes can be measured from eyes (EOG) changes blood volume (plethysmograph). sorted plethysmograph, (hEOG + vEOG), vEOG, hEOG, zEMG, tEMG, EMG (tEMG zEMG), respiration, EDA, temperature plethysmograph. Originality/value Many studies literature dependent limited report an average leave out (LOSO) validation result accuracy. work reported sets baseline clearly specifying subjects training set. In addition, specifies cut-off score classify scale low or high dimensions. Generally, statistical a modality, whereas work, used. identified predicted