Abstract:The additive partially linear model (APLM) combines the flexibility of nonparametric regression with the parsimony of regression models, and has been widely used as a popular tool in multivariate nonparametric regression to alleviate the "curse of dimensionality". A natural question raised in practice is the choice of structure in the nonparametric part, that is, whether the continuous covariates enter into the model in linear or nonparametric form. In this paper, we present a comprehensive framework for simultaneous sparse model identification and learning for ultra-high-dimensional APLMs where both the linear and nonparametric components are possibly larger than the sample size. We propose a fast and efficient two-stage procedure. In the first stage, we decompose the nonparametric functions into a linear part and a nonlinear part. The nonlinear functions are approximated by constant spline bases, and a triple penalization procedure is proposed to select nonzero components using adaptive group LASSO. In the second stage, we refit data with selected covariates using higher order polynomial splines, and apply spline-backfitted local-linear smoothing to obtain asymptotic normality for the estimators. The procedure is shown to be consistent for model structure identification. It can identify zero, linear, and nonlinear components correctly and efficiently. Inference can be made on both linear coefficients and nonparametric functions. We conduct simulation studies to evaluate the performance of the method and apply the proposed method to a dataset on the Shoot Apical Meristem (SAM) of maize genotypes for illustration.

NONLINEAR STRUCTURE ANALYSIS WITH PARTIAL LEAST-SQUARES REGRESSION BASED ON SPLINE TRANSFORMATION

Least squares support vector machine based partially linear model identification

Using Svm To Model And Control Nonlinear Dynamical Systems

Polynomial spline estimation for partial functional linear regression models

ADDITIVE COEFFICIENT MODELING VIA POLYNOMIAL SPLINE

Partial Linear Models for Longitudinal Data Based on Penalized General Method of Moments

Nonlinear Projection to Latent Structures Method and Its Applications

Identification of partially linear structure in additive models with an application to gene expression prediction from sequences

Spline-Backfitted Kernel Smoothing Of Nonlinear Additive Autoregression Model

Polynomial spline regression: Theory and Application

Sparse Model Identification and Learning for Ultra-high-dimensional Additive Partially Linear Models

Physics-informed Spline Learning for Nonlinear Dynamics Discovery

Penalized Spline Estimation for Partially Linear Single-Index Models

Spline based sparseness and smoothness for partially nonlinear model via C-fused Lasso

Penalized Partial Least Squares Based on B-Splines Transformations

Estimating Functions and Derivatives Via Adaptive Penalized Splines.

Sparse and Efficient Estimation for Partial Spline Models with Increasing Dimension

Mapping-to-Parameter Nonlinear Functional Regression with Novel B-spline Free Knot Placement Algorithm

Spline Estimation of Single-Index Models

A robust spline approach in partially linear additive models

Spline-Backfitted Kernel Smoothing of Partially Linear Additive Model