论文部分内容阅读
针对不确定数据多分类问题,提出一种基于模糊机会约束的超球支持向量机(FCC-HSVM).在球结构支持向量机的基础上,引入模糊事件的可能性测度,得到模糊机会约束规划及其对偶规划.利用球结构的优点,每类样本只参与一个超球体的训练,直接求解多分类模型,提出FCC-HSVM的快速学习算法,显著缩短多分类情况下训练时间.数据试验表明:这种支持向量机分类精度较高,训练速度快,适合解决不确定数据多分类问题.
In order to solve the problem of multi-classification of uncertain data, a new hypersphere support vector machine (FCC-HSVM) based on fuzzy chance constrains is proposed. On the basis of ball-structure support vector machine, the fuzzy probability constraint programming And their dual programming.Using the advantages of the ball structure, each type of sample only involved in the training of a hypersphere, directly solve the multi-classification model, proposed FCC-HSVM fast learning algorithm, significantly reducing the training time under multi-classification.Data experiments show that: This support vector machine classification accuracy, training speed, suitable for solving multi-classification of uncertain data problems.