模糊支持向量机情感状态识别的研究*

针对已有的情感生理参数样本类内聚合度低、不同状态较难区分的特点,提出了一种改进的模糊支持向量机识别方法.模糊隶属度函数采用高斯分布形式,高斯分布的参数分别由同类样本数据形成的最小超球体半径和样本之间的紧密程度决定.该方法计算样本模糊隶属度时,不仅考虑样本与类中心的距离关系,还要考虑样本与样本之间的关系.实验显示改进的模...

Study of fuzzy support vector machine emotional state recognition

Physical parameters of the human objectively reflect emotional states, Many research studies in this area. Because the Emotional physiological mechanism is not clear, it is very difficult to accurately determine emotional state. Physiological parameters of emotional low degree of polymerization, emotional state support vector machine Training large computation and low accuracy. Fuzzy Evaluation on the training samples based on the classification, reducing ordinary data and the impact of outliers on training to improve physiological parameters on the emotion recognition accuracy. Experiments show that the incremental learning algorithm for fuzzy identification is obvious. Sample assessment improves support vector machines recognition accuracy.