非线性支持向量分类机解的二阶充分条件  

Second Order Sufficient Conditions Property for Support Vector Classification

作　　者：蔡春[1] 吕书强[1] 徐坚[2] 

机构地区：[1]北京联合大学应用文理学院,北京100191 [2]北京联合大学应用科技学院,北京100191

出　　处：《数学的实践与认识》2015年第3期264-267,共4页Mathematics in Practice and Theory

基　　金：北京市属高等学校高层次人才引进与培养计划项目;CIT&TCD201404080;北京联合大学新起点计划项目资助~~

摘　　要：优化问题解的二阶充分条件是研究其灵敏度分析的基础,支持向量分类机是新的数据挖掘优化问题.给出了支持向量分类机的解满足二阶充分条件成立定理;定理的假设条件是很弱的,用支持向量分类机求解实际问题,通常总假定这一条件成立;特别地,对线性可分支持向量机问题,其解满足二阶充分条件成为当然成立的事实.Second order sufficient condition is the foundation for optimal problem sensitivity analysis. Support vector classification is new data mining optimal model. The paper presents one hypothesis which guarantees that second order sufficient condition property of support vector classification holds. The hypothesis is much weak. One reason is that the hypothesis holds certainly for linearly separable support vector classification. In addition, the linear support vector classification solution is solved under such a hypothesis.

关 键 词：支持向量分类机 二阶充分条件 灵敏度分析 数据挖掘 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]