A note on penalized minimum distance estimation in nonparametric regression

被引：2

作者：

Bunea, F ^{[1
]}

Wegkamp, MH ^{[1
]}

机构：

[1] Florida State Univ, Dept Stat, Tallahassee, FL 32306 USA

来源：

CANADIAN JOURNAL OF STATISTICS-REVUE CANADIENNE DE STATISTIQUE | 2003年 / 31卷 / 03期

关键词：

complexity penalty; L-1-risk; minimum distance estimators; model selection; non-parametric regression; shatter coefficient;

D O I：

10.2307/3316086

中图分类号：

O21 [概率论与数理统计]; C8 [统计学];

学科分类号：

020208 ; 070103 ; 0714 ;

摘要：

The authors introduce a penalized minimum distance regression estimator. They show the estimator to balance, among a sequence of nested models of increasing complexity, the L-1-approximation error of each model class and a penalty term which reflects the richness of each model and serves as a upper bound for the estimation error.

引用

页码：267 / 274

页数：8

共 12 条

[1]

Anthony M., 1999, Neural network learning: theoretical foundations, Vfirst

[2]

BIAU G, 2002, UNPUB NOTE DENSITY

[3]

BIAU G, 2002, UNPUB DENSITY ESTIMA

[4]

Devroye L, 1996, ANN STAT, V24, P2499

[5]

Devroye L, 1997, ANN STAT, V25, P2626

[6]

DEVROYE L, 2000, COMBINATORIAL METHOD

[7]

Devroye L., 1996, A probabilistic theory of pattern recognition

[8]

Gyorfi L., 2002, A Distribution-Free Theory of Nonparametric Regression, V1

[9] Estimation and selection procedures in regression:: an L1 approach [J].

Hengartner, N ;

Wegkamp, M .

CANADIAN JOURNAL OF STATISTICS-REVUE CANADIENNE DE STATISTIQUE, 2001, 29 (04) :621-632

[10]

Nicoleris T, 1997, ANN STAT, V25, P2493

← 1 2 →