“一对多”支持向量机多类分类方法缺点就是一类对余类的这些两类问题是很不对称的,在样本训练中,数目少的类别往往比数目多的类别训练误差更大。 “One to many ” Support vector machine Multi-class classification method The disadvantage is that a class of these two types of problems for the remaining class is very asymmetric, in the sample training, a small number of categories tend to be more than the number of categories of training errors more Big.