摘要
为了改进半参数模型补偿最小二乘估计中的正则化参数选取方法,该文设计了3种方案:L-曲线法、虚拟观测法以及该文提出的改进方法以比较正则化参数的优劣。模拟算例表明,该文提出的改进方法可以有效评价正则化参数,得到的参数估值精度更高,是一种较为有效的选取补偿最小二乘最佳正则化参数的方法。
In order to improv the selection method of regularized parameter inthe penalized least squares of semi-parametric model,this paper designed three programs：L-curve,virtual observation method and the proposed method of the improvement to compare regularized parameter.The simulation example showed that the proposed method could evaluate regularized parameter effectively and obtain higher accuracy of parameter estimated value,which was an relatively effective method of selecting best penalized least squares regularized parameter.
作者
周敏
周世健
谯婷
池其才
ZHOUMin;ZHOUShijian;QIAO Ting;CHI Qicai(Faculty of Geomatics, East China University of Technology, Nanchang 330013, China;Guangzhou Okay Information Technology Co. Ltd. , Guangzhou 510663, China;Key Laboratory of Watershed Ecology and Geographical Environment Monitoring, NASG, Nanchang 330013, China;Nanchang Hangkong University, Nanchang 330063, China;Real Estate Surveying and Mapping Assessment Office of Nansha District of Guangzhou, Guan- gzhou 511458, China)
出处
《测绘科学》
CSCD
北大核心
2018年第4期105-108,共4页
Science of Surveying and Mapping
基金
国家自然科学基金项目(41374007)
东华理工大学研究生创新基金项目(DHYC-2016019)
测绘地理信息江西省研究生创新教育基地项目
关键词
半参数模型
正则化参数
模型误差
补偿最小二乘
semi-parametric model
regularized parameter
model error
penalized least squares
