Abstract:
Emotion is an important aspect in the interaction between humans. It is fundamental to human experience and rational decision-making. There is a great interest for detecting emotions automatically. A number of techniques have been employed for this purpose using channels such as voice and facial expressions. However, these channels are not very accurate because they can be affected by users' intentions. Other techniques use physiological signals along with electroencephalography (EEG) for emotion detection. However, these approaches are not very practical for real time applications because they ask the participants to reduce any motion and facial muscle movement, reject EEG data contaminated with artifacts and rely on large number of electrodes. In this thesis, we propose an approach that analyzes highly contaminated EEG data produced from a new emotion elicitation technique. We also use a feature selection mechanism to extract features that are relevant to the emotion detection task based on neuroscience findings. We reached an average accuracy of 51% for joy emotion, 53% for anger, 58% for fear and 61% for sadness. We are also, applying our approach on smaller number of electrodes that ranges from 4 up to 25 electrodes and we reached an average classification accuracy of 33% for joy emotion, 38% for anger, 33% for fear and 37.5% for sadness using 4 or 6 electrodes only.
