半参数模型中影响正则化参数的因素分析

¹中南大学地球科学与信息物理学院,长沙市麓山南路932号410083;²湖南省普通高等学校精密工程测量与形变灾害监测重点实验室,长沙市麓山南路932号410083

基金项目: 国家自然科学基金资助项目(40974007);;湖南省教育厅基金资助项目(09K006)

详细信息

Analysis of Factors Influencing Smoothing Parameter in Semiparametric Model

¹ School of Geosciences and Info-physics,Central South University,932 South Lushan Road,Changsha 410083,China;² Key Laboratory of Precise Engineering Surveying & Deformation Hazard Monitoring,Universities of Hunan Province, 932 South Lushan Road,Changsha 410083,China

Funds: 国家自然科学基金资助项目(40974007);;湖南省教育厅基金资助项目(09K006)

摘要: 通过模拟算例,比较了L曲线法、GCV法(广义交叉核实法)和虚拟观测法确定出的正则化参数,并对影响正则化参数的因素进行了分析,得出结论:正则化参数的确定与信噪比密切相关,当信噪比增大时,各种方法确定的正则化参数变化趋势不同,不同的情况确定正则化参数的适用方法也会有所差异。
- 正则化参数 /
- 影响因素 /
- L曲线法 /
- GCV法 /
- 虚拟观测法
Abstract: Many methods can be used to determine smoothing parameters in semiparametric model,but smoothing parameters vary with these methods for the same model.After comparative analysis of the L-Curve,GCV(Generalize Cross Validation) and virtual observational approach,the factors influencing smoothing parameter are studied by simulated example.Smoothing parameters have a close relation with signal-to-noise ratio.When the signal-to-noise ratio increases,variation trend of smoothing parameters is different.The best method determining smoothing parameters in different situations is different.
- smoothing parameters /
- influencing factors /
- L-Curve /
- GCV /
- virtual observation