摘要
支持向量回归超参数的选择会影响模型性能,常用的梯度下降选择方法要求核函数或估计函数近似可微,且对迭代初值具有较强依赖性.对此,给出一种两阶段参数优化选择方法.第1阶段根据问题实际需求,确定超参数的变化区域;第2阶段在确定的参数变化范围内,采用自适应混沌文化算法,寻找具有最优性能的超参数组合.面向Mackey-Glass时间序列预测的仿真结果表明,该参数选择方法对函数结构不具有依赖性,所得超参数对应的SVR模型具有较好的泛化性能.
The hyper-parameters of support vector regression influence the performance of its model.In the normal gradient descent method,kernel functions or estimation functions must approximately differential,and this method sensitively depends on initial values.Therefore,a two-stage optimization selection method for hyper-parameters is proposed.In first stage,the search extent of each hyper-parameter is determined according to the reqirement of issues.In second stage,optimal hyper-parameters are obtained by using adaptive chaotic culture algorithm during above search space.Taken the prediction of Mackey-Glass time series as example,simulation results show that the selection method is not related on the stucture of functions,and support vector regression(SVR) model corresponding to optimal hyper-parameters by using this method has better generaliztion.
出处
《控制与决策》
EI
CSCD
北大核心
2010年第4期525-530,共6页
Control and Decision
基金
国家自然科学基金项目(60805025)
中国博士后基金项目(20090460328)
江苏省青蓝工程
关键词
支持向量回归
超参数
两阶段
自适应混沌文化算法
时间序列
Support vector regression
Hyper-parameters
Two-stage
Adaptive chaotic cultural algorithm
Time series