EEG Based Emotion Detection Using Machine Learning Models

This abstract highlight that emotion plays a huge role in a human’s daily interaction. Currently, emotion recognition systems are very useful in different areas from human computer interaction (HCI) systems to the issues of healthcare safety. Emotion recognition is a highly interesting field of research at the boundary between neuroscience and artificial intelligence, which attempts to obtain information about human affective states from spontaneous EEG signals. This research leverages machine learning approaches in EEG based emotion recognition to overcome some important gaps about the task. Thus far, the Electroencephalogram (EEG) is most widely investigated due to its sensitivity to changes in the mood. But the relationship has not been fully researched. In order to investigate the relationship between emotions, bands, channels and then suggest a better classifier, this experiment organizes the framework with the EEG dataset. Following the data preprocessing, EEG data is decreased by principal component analysis (PCA). The parameter chosen to then create the emotion feature vectors is discrete wavelet transform (DWT). In this regard, here a Convolution Neural Network is proposed for EEG signal classification and compared with Support Vector Machine (SVM) and Random Forest (RF).