一种新的模糊支持向量机方法

针对分类问题的支持向量机方法中存在对噪声和野值敏感的问题，提出了一种新的隶属度函数确定方法．研究和分析了目前针对这些问题所提出的模糊支持向量机方法，而在模糊支持向量机方法中，隶属度函数的设计是整个模糊算法的关键所在，文中主要对隶属度函数进行了深入地研究，在类中心向量方法的基础上，即在确定样本的隶属度时，对类中心距离方法进行改进，得到新的隶属度计算的方法，并从数值实验方面表明该方法的有效性和可行性．

A New Fuzzy Supportive Vector Machine Algorithm

SVM is very sensitive to noises and outliers when solving the problem of classification, a new method of membership function is put forward. This paper studies and analyzes the present Fuzzy Supportive Vector Machine method in view of these questions. In these methods, the design of membership function is a key to solve the classification problem with FSVM. This paper mainly studies membership function. Based on the center vector method, a new membership method is provided. This approach greatly improves the classification accuracy and generalization. Experiment results have confirmed its effectiveness and feasibility.