Abstract
EEG signals reflect the inner emotional state of a person and regarding its wealth in temporal resolution, it can be used profitably to measure mental stress. Emotional states recognition is a growing research field inasmuch to its importance in Human-machine applications in all domains, in particular psychology and psychiatry. The main goal of this study is to provide a simple method for stress detection based on Frontal Alpha Asymmetry for trials selection and time, time-frequency domain features. This approach was tested on prevalent DEAP database, and provided us with two subdatasets to be processed and classified thereafter. From the variety of features produced in the literature we chose to test Hjorth parameters and Band Power as a time-frequency feature. To enhance the classification performance, we tested the SVM classifier, K-NN and Fuzzy K-NN.