ELECTROCARDIOGRAM-BASED MULTI-CLASS SVM CLASSIFIER FOR EMOTION RECOGNITION
The ability to recognize emotion is one of the hallmarks of emotional intelligence. In this paper, we discuss the most important stage of a fully implemented emotion recognition system including data analysis and classification. For collecting electrocardiogram (ECG) signals in different affective states, we use a video induction method which elicits naturally emotional reactions from the multiple subjects. A multi-class support vector machine is adopted as a pattern classifier to recognize the subjects’ emotion. After calculating a sufficient amount of features from the raw signals, principal component analyses based feature selection/reduction method is tested to extract a new feature set consisting of the most significant features for improving classification performance. The average recognition sensitivity on four classes of emotion, including joy, anger, sadness, and pleasure, is enhanced from 76.7％ to 82.5％.
Emotion recognition ECG signal processing Multi-class SVM Principal component analyses
Bo-Hui Zhu Li-Hong Ren Yong-Sheng Ding Yi-Zhi Wu Kuang-Rong Hao
College of Information Sciences and Technology,Donghua University,Shanghai 201620,P.R.China College of Information Sciences and Technology,Donghua University,Shanghai 201620,P.R.China;Engineer
国际会议
2008高等智能国际会议(2008 International Conference on Advanced Intelligence)
北京
英文
2008-10-18（万方平台首次上网日期，不代表论文的发表时间）