A Pyramidal Approach for Emotion Recognition from EEG Signals

Abstract

Brain Computer Interfaces (BCI) is one of the key technology gaining intense mode of interest in various fields of research in artificial intelligence. In recent years, physiological signals with advanced BCI applications, successively concerned in the direction of recognizing various human emotional states. In this paper, a new feature representation technique based on pyramidal approach is determined. The proposed approach uses the Interpolation Forward Difference signal computations in different levels of iterations, which leads to reduction of dimensions in an effective way for recognizing emotions. Then application of General Regression Neural Network (GRNN) is presented for efficient classification of signals into four different emotional states from EEG based GAMEEMO Dataset. The experimental results are promising and performed well, compared to other state of art techniques.