Few studies have focused on integrating affective computing and soft computing technique for human physiology signals recognition in digital content learning system. This study develops a human affective norm (emotion and attention) recognition system for U-learning system. Eight elementary school students (4 male and 4 female) were recruited as participants to see some emotion pictures in international affective picture system (IAPS), and to collect their affective information-attention, meditation, electroencephalography (EEG), electrocardiogram (ECG), and SpO2 for developing the affective norm recognition system. These bio-physiology signals were extracted important features selection (Factor analysis) to serve as the input variables for radial basis function (RBF) neural network model. The results showed that all types of factor analysis did not perform well in our emotion norm database. Factor analysis with covariance extraction has higher accumulative variances than correlation extraction. This study suggested that future research can adopt more nonlinear feature selection methods to develop a high accuracy support vector machine based emotion recognition system.
- Factor Analysis as the Feature Selection Method in an Emotion Norm Database
- Chih-Hung Wu - Natl Taichung Univ Educ, Taichung, TaiwanBor-Chen Kuo - Natl Taichung Univ Educ, Taichung, TaiwanGwo-Hshiung Tzeng - University of Taipei
- N T Nguyen (Editor)B Attachoo (Editor)B Trawinski (Editor)K Somboonviwat (Editor)
- Lecture Notes in Artificial Intelligence
- Springer Nature
- 10
- Conference proceeding
- 01/01/2014
- INTELLIGENT INFORMATION AND DATABASE SYSTEMS, PT II, Vol.8398(2), pp.332-341
- English