EEG Bases Emotion Detection Using Deep Learning Algorithm

Human computer interaction is a fast growing area of research where in the physiological signals are used to identify human emotion states. Now a day, gaming Disorder has become a major source of concern in recent years and has therefore created immense interest among researchers for further study. In this paper, work is carried out to detect the emotional behavior of the subject who involve in online games regularly using Electroencephalography (EEG). EEG is a low cost, high temporal resolution popular tool which can be used for studying addictive behaviors. As a result, EEG will serve as a reliable indicator of the subject's emotional state. In the present work, the training model is for emotion states is done using SEED-IV data base and testing of gamming addiction behavior is done for acquired signals. The spectrogram features are fed to the VGG pre-trained model. The trained model performance of prediction accuracy of 89.54% and testing accuracy of 78.63% on SEED-IV database is obtained. The acquired signals are tested on the trained model and an accuracy of 75% is obtained.