摘要
利用再生核空间方法研究相依误差近似回归模型的稳健设计问题.假定模型偏差来自概率测度为P的函数类H,同时还假定不同试验所产生的误差不独立.采用广义最小二乘估计作为参数向量的估计,根据平均预测风险最小原理得到了一个稳健设计准则.数值例子表明由该准则得到的设计具有良好的稳健性.
This paper studies the model - robust design problem by applying the reproducing kernel space approach. We assume that the model has an unknown bias or contamination from some class (?) with a probability measure P, and the correlated errors are considered. We develop a design criterion in terms of the average expected quadratic loss for generalized least squares estimation. Numerical results indicate the designs obtained through this criterion are more robust.
出处
《上海师范大学学报(自然科学版)》
2004年第3期16-22,共7页
Journal of Shanghai Normal University(Natural Sciences)
基金
ThisworkispartiallysupportedbytheNationalScienceFoundationofChinagrant10271078theSpecialFundsforMajorSpecialtiesoftheShanghaiEducationCommitteeandShanghaiNormalUniversity.