Emotion recognition model using source-temporal features and fuzzy

• Main Authors: Shams, Wafa Khazal, Abdul Rahman, Abdul Wahab, Taha Alshaikhli, Imad Fakhri
• Format: Article
• Language: English
• Published: Springer-Verlag, Berlin, Germany 2013
• Subjects: QA75 Electronic computers. Computer science
• Online Access: http://irep.iium.edu.my/34356/1/Emotion_Recognition_Model_Using_Source-Temporal_Features_and_Fuzzy.pdf; http://irep.iium.edu.my/34356/; http://www.springer.com/computer/lncs?SGWID=0-164-6-1068921-0

This paper aims to evaluate the performance of Electroencephalographic (EEG) emotion recognition system (EEG-ER) using source-temporal domain with Takagi- Sugeno-Kang (TSK) fuzzy model. Ten healthy subjects aged 5-6 years were participated in this study. Emotion elicitation procedure has done using the Radbound faces database (RafD). The selected emotions were happy, sad, and neutral and fear. The results were compared with wavelet coefficients (WC) as feature extraction method and Regularized Least Square (RLS) and Multi-Layer Perception (MLP) neural network classifiers from our previous work. Another comparison was done between affective model of Russell and RafD. The results show the efficiency of using source-temporal features in emotion recognition system hence there was a slight difference in accuracy among different classifier; MLP, RLS and TSK however MLP and TSK results were with high accurate and stable. Moreover Russell model which is based on positive-negative dimensions shows high accuracy than RafD model that has positive dimensions. The accuracy was around 97% using Russell model.