摘要: |
优化问题的二阶充分条件是研究灵敏度分析的基础,支持向量机是数据挖掘的新方法。针对线性支持向量分类机优化问题,研究了其解的二阶充分条件,给出了二阶充分条件成立的假设条件。研究表明,该假设条件很弱,用支持向量机算法求解实际问题时,通常假定这一条件成立,特别地,对线性可分支持向量分类机优化问题,其解一定满足这一条件,满足二阶充分条件成为当然成立的事实。 |
关键词: 支持向量分类机,二阶充分条件,数据挖掘,起作用约束 |
DOI:10.11841/j.issn.1007-4333.2006.06.161 |
投稿时间:2006-09-15 |
基金项目:国家自然科学基金资助项目(10371131) |
|
Second order sufficient condition for optimization problem of a linear support vector classifier |
|
|
Abstract: |
A second-order-sufficient-condition is one of basis for sensitivity analysis of optimization problems.A hypothesis to ensure the tenability of a second-order-sufficient-condition of a linear support vector classifier is presented in the paper,which is a weak one for the solution of a linearly separable support vector classifier must meet it.In addition,the problem of a linear support vector classifier is solved under such a hypothesis. |
Key words: support vector classifier,second order sufficient condition,data mining,active constraint, |