Determining Positive-Negative Emotions in Male and Female Based on EEG Signals using Machine Learning Algorithms

Emotionsare vital in everyday human life as a controller of behavior, decision-making and as a means to determine product marketing strategy/ market research. All of these things are very dependent on human emotional conditions. In addition, in the development of the computational affective field, brain signal-based emotion recognition (EEG) has become a trending topic of current research. Therefore, we attempted to compare positive-negative emotions in men and women based on EEG using a Machine Learning algorithm in this study. A total of 20 male and 20 female participants recorded their EEG signals in the frontopolar and frontal areas of the brain. Then the EEG data is processed by filtering, removing artifact, and decomposing it into three sub-bands (alpha, beta, and gamma). The extracted signal features are Mean Absolute Deviation and Power Spectral Density. Based on the signal feature analysis results, it is known that the signal feature values (MAD and PSD) for women tend to be higher than for men. Meanwhile, several algorithms are used to classify positive and negative emotions, such as Naive Bayes, K-Nearest Neighbor, Support Vector Machine, and Random Forest. Based on the results of classification, the best accuracy rate was 95.8% (on positive emotions for male & female gender), 92.2% (on negative emotions for male & female gender), and 79.8% (on positive-negative emotions for male & female gender) using Random Forest algorithm.