Improving emotion detection through artificial intelligence from EEG brainwave signals

The detection of human mental states has applications in a variety of fields, including healthcare, robotics, neurology, etc. Fundamental human emotion can be identified by facial expressions or bodily movements, but when we need to evaluate the emotions specifically and more precisely. Electroencephalogram (EEG) neuroimaging and other techniques are still in the early stages of detecting a range of emotions shown by individuals. With this brain wave analysis, we will be able to comprehend a respondent’s true feelings, even if they are trying to hide them. The purpose of this study was to use EEG brain wave signals for the detection of emotions and to classify them into three mental states relax, neutral, and concentrating with the help of different artificial intelligence models. The publicly available dataset of the Muse headband was used which was comprised of EEG brainwave signals from four EEG sensors (AF7, AF8, TP9, TP10). As evaluators, we used machine learning models such as Nave Bayes, Bayes Net, J48, Random Tree, and Random Forest, as well as feature selection methods: OneR, information gain, correlation, and symmetrical uncertainty. The overall accuracy of Random Forest was better (95.07%) as compared to other models.