Dysphoria recognition based on EEG / Nur Aqilah Husna Anuar

Dysphoria in general, can be defined as an affective state that is related to negative emotions whereas emotion is the observation of mental states. The examples of negative emotion are sad, fear and anger. Dysphoria may lead to depression and if untreated, it can result to severe mental health issues such as anxiety, bipolar, depression and schizophrenia, and acts as a catalyst to suicide and drug abuse. However, dysphoria is difficult to be detected because there is no empirical measurement to detect dysphoria making it almost difficult to achieve an optimal diagnosis. The purpose of this project is to provide an empirical measurement that can be used to detect dysphoria. The brain signals are captured using an electroencephalogram (EEG) to study the negative emotions related to dysphoria. Electroencephalogram (EEG) is the recording of the electric activity generated by the brain. The main objective of this project is to study dysphoria related features captured using electroencephalogram (EEG) and classified using multiple classifiers. The performances of the classifiers are then compared. The Mel-Frequency Cepstral Coefficient (MFCC) method is used to extract relevant feature and three classifier which are Multi-Layer Perceptron (MLP), Naïve Bayes (NB) and Random Forest (RF) are employed for classification. The accuracy of the three classifiers are compared to determine which classifier perform the best. Based on the experimental results, MLP perform better as compared to NB and RF. The finding indicated that emotion can be recognized and the tendency for stress identification is feasible.