一种基于粗糙间隔的模糊支持向量机

A Rough Margin Based Fuzzy Support Vector Machine

以模糊支持向量机(FSVM)为基础,同时考虑样本在间隔中的位置对决策超平面的影响,提出了基于粗糙间隔的模糊支持向量机(RFSVM).通过计算各个数据点的模糊隶属度,并利用最大化粗糙间隔方法,对具有隶属度的数据进行训练以获得决策超平面.在此算法中,位于下间隔中的训练点比边界域中的训练点具有较大的惩罚值,以便更好地减少噪声或野点对超平面的影响.利用选择的标准数据集对几种不同算法进行了实验比较,结果表明了RFSVM算法的有效性.

Based on fuzzy support vector machine（FSVM）,we presented a rough margin based fuzzy support vector machine （RFSVM）by introducing the effects of positions of training samples in the margin on decision hyper-plane in this paper.After computing the degree of fuzzy membership of each training point,we used these data for training to obtain the decision hyper-plane by maximizing rough margin＇s method.In this algorithm,points in the lower margin have major penalty than those in the boundary.We compared RFSVM with other support vector machine algorithms on several benchmark datasets.Experimental results show that RFSVM is effective and feasible.