An Approach to EEG Based Emotion Recognition and Classification Using Kernel Density Estimation

This paper aims to proposed emotion recognition using electroencephalography (EEG) techniques. Recognizing emotion by using computers is becoming popular these days. This paper is based on calculating EEG signals and recognizing emotion from human brain activity. Electroencephalogram (EEG) signals are taken from the scalp of the brain and assessed in responds to several stimuli from the four basic emotions on the IAPS emotion stimuli. Features from the EEG signals are captured using the Kernel Density Estimation (KDE) and classified via the artificial neural network classifier to recognise emotional condition of the subject under test. Results are obtained to prove that the proposed modified KDE gives better results in terms of accuracy. Also, the proposed method gives better estimation of emotion of the subject from streaming EEG data by using the concept of cluster kernels.