Emotion Recognition via Detrended Fluctuation Analysis and Fractal Dimensions

Although there are several methods that have been used in emotion recognition, still the ultimate goals of these researches are the same. One of these goals is to find the suitable features required for emotion recognition using different analytic techniques besides an accurate classification model to improve the accuracy of emotion classification system. This study has proposed an automatic emotion recognition system based on three bio signals EEG, EMG, and GSR using a combination of Detrended Fluctuation Analysis (DFA) and four fractal dimensions. Three supervised classifiers have been used to classify these signals into three emotional classes along the two emotional dimensions, the maximum achieved accuracies were 94.3% for valence dimension and 94% for arousal dimension.