基于改进FSVM的数据挖掘分类算法

Data mining classification algorithm based on improved fuzzy support vector machine

针对模糊支持向量机(FSVM)应用于数据挖掘分类中存在对大样本集训练速度慢以及对噪声点敏感影响分类正确率的问题,提出一种基于改进FSVM的数据挖掘分类算法.该算法首先预选有效的候选支持向量,减小训练样本数目,提高训练速度;其次定义一种新的隶属度函数,增强支持向量对构建模糊支持向量机的作用;最后将近邻样本密度应用于隶属度函数设计,降低噪声点或野值点对分类的影响提高分类正确率.实验结果表明,该算法在训练样本数目较大时训练速度和分类正确率都有提高.

Aimed at the problem in data mining classification with fuzzy support vector machine（FSVM）such as slow training speed of big sample data-sets and sensitivity to noises that affects the validity of classification,an improved FSVM-based method for data mining classification is proposed.First,in this algorithm the effective candidate support vectors are preselected to reduce the number of training samples to improve training speed.Second,a novel membership function is defined to enhance the function of support vectors in construction of FSVM.Finally,the neighborhood sample density is applied to the design of membership function to reduce the influence of the noises or outliers on the classification to improve classification validity.Experimental results show that the proposed algorithm will make the training speed and classification validity improved when the number of the training samples are bigger.