摘要
v-支持向量回归机(v-support vector regression,v-SVR)的对偶形式与ε-支持向量回归机的对偶形式相比增加了一个额外的不等式约束,截止目前还没有找到有效且可行的v-SVR的v解路径算法。针对Loosli等人提出的v-SVR的v解路径算法存在路径不可更新的问题,提出了改进的v-SVR的v解路径算法。该算法基于v-SVR的修改形式及Karush-Kuhn-Tucker(KKT)条件,通过引入新的变量和附加项的策略,能够有效地避免在绝缘增量调整过程中存在的冲突和异常,并最终经过有限次数迭代拟合出整个v解路径。理论分析和仿真结果表明,该算法是有效且可行的。
In comparison with the dual formulation ofε-support vector machine,the dual of v-support vector regression（v-SVR）has an extra inequality constraint.To date,there is no effective and feasible vsolution path for v-SVR.To solve the infeasible updating path problem of the vsolution path for v-SVR,which was first proposed by Loosli et al,an improved vsolution path for v-SVR is proposed.Based on the modified formulation of v-SVR and the Karush-Kuhn-Tucker（KKT）conditions,the strategy of using a new introduced variable and an extra term can avoid the conflicts and exceptions effectively during the adiabatic incremental adjustments.Finally,the proposed algorithm can fit the entire vsolution path within the finite number of iterations.Theoretical analysis and simulation results demonstrate that the proposed algorithm is effective and feasible.
出处
《系统工程与电子技术》
EI
CSCD
北大核心
2016年第1期205-214,共10页
Systems Engineering and Electronics
基金
国家自然科学基金(61273131)
江苏省产学研联合创新资金项目(BY2013015-39)资助课题
关键词
机器学习
模型选择
v-支持向量回归机
v解路径
machine learning
model selection
v-support vector regression（v-SVR）
vsolution path
