An RKHS-Based Semiparametric Approach to Nonlinear Sufficient Dimension Reduction

TL;DR

This paper introduces a novel RKHS-based semiparametric method for nonlinear sufficient dimension reduction that handles infinite-dimensional parameters without relying on traditional linearity or constant variance assumptions.

Contribution

It extends semiparametric dimension reduction to a generalized RKHS framework, deriving estimators and their asymptotic properties for nonlinear SDR.

Findings

The method effectively estimates dimension reduction directions.

It outperforms existing methods in simulations and real data.

The approach does not depend on linearity or constant variance conditions.

Abstract

Based on the theory of reproducing kernel Hilbert space (RKHS) and semiparametric method, we propose a new approach to nonlinear dimension reduction. The method extends the semiparametric method into a more generalized domain where both the interested parameters and nuisance parameters to be infinite dimensional. By casting the nonlinear dimensional reduction problem in a generalized semiparametric framework, we calculate the orthogonal complement space of generalized nuisance tangent space to derive the estimating equation. Solving the estimating equation by the theory of RKHS and regularization, we obtain the estimation of dimension reduction directions of the sufficient dimension reduction (SDR) subspace and also show the asymptotic property of estimator. Furthermore, the proposed method does not rely on the linearity condition and constant variance condition. Simulation and real data studies are conducted to demonstrate the finite sample performance of our method in comparison with several existing methods.