模糊支持向量机情感状态识别的研究  

Study of fuzzy support vector machine emotional state recognition

作　　者：徐鲁强[1,2] 刘静霞[3] 

机构地区：[1]西南科技大学计算机学院,四川绵阳621010 [2]西南交通大学信息科学与技术学院,成都610000 [3]成都电子机械高等专科学校,成都610031

出　　处：《计算机应用研究》2011年第3期831-832,837,共3页Application Research of Computers

基　　金：国家自然科学基金资助项目(60572143)

摘　　要：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。实验显示改进的模糊支持向量机方法识别性能得到提高。Due to low similarity for the same emotional state parameters and difficult to distinguish between different emotional states,this paper proposed an improved fuzzy support vector machine recognition method.Fuzzy membership function took the form of the Gaussian function,determined Gaussian function parameters by the radius of the same state sample data smallest hyper sphere and tightness of sample data.Determined fuzzy membership value of sample by not only considered the distance between the sample classes and the center of the sample the class,but also considered the relationship betwwen samples.Experiments show that the improved fuzzy support vector machine recognition performance is improved.

关 键 词：情感状态识别 模糊支持向量机 情感生理参数 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]