改进的推理型支持向量机  

The Improvement of Transductive Support Vector Machine

作　　者：张庆[1] 阎满富[1] 

机构地区：[1]唐山师范学院数信系,河北唐山063000

出　　处：《数学的实践与认识》2008年第9期136-140,共5页Mathematics in Practice and Theory

基　　金：国家自然科学基金重点项目(10631070);国家自然科学基金面上项目(10601064)

摘　　要：自V apn ik于20世纪90年代末提出推理型支持向量机的概念后,关于推理型支持向量机的研究基本处于停止状态,主要问题是这种支持向量机的优化模型求解有相当的困难.文章试图把它的优化问题变为无约束问题,再构造带有核的光滑无约束最优化问题,由此构建最优化问题易于求解的推理型支持向量机,以突破对它深入研究的瓶颈.Study on Transductive Support Vector Machine （TSVM） has made little progress since late 1990s when Vapnik put forth such concept, as algorithm for TSVM optimization model can not be easily found. Here the paper try to transform the problem of TSVM optimization into an unconstraint one before constructed smoothing unconstraint optimization that has a kernel, and on the basis of which to devise a TSVM whose optimization problem is easier to solve to break through the bottleneck in order to deepen the research into TSVM.

关 键 词：最优化 光滑函数 无约束问题 推理型支持向量机 

分 类 号：O234[理学—运筹学与控制论]