Abstract:Sliced inverse regression (SIR, Li 1991) is a pioneering work and the most recognized method in sufficient dimension reduction. While promising progress has been made in theory and methods of high-dimensional SIR, two remaining challenges are still nagging high-dimensional multivariate applications. First, choosing the number of slices in SIR is a difficult problem, and it depends on the sample size, the distribution of variables, and other practical considerations. Second, the extension of SIR from univariate response to multivariate is not trivial. Targeting at the same dimension reduction subspace as SIR, we propose a new slicing-free method that provides a unified solution to sufficient dimension reduction with high-dimensional covariates and univariate or multivariate response. We achieve this by adopting the recently developed martingale difference divergence matrix (MDDM, Lee & Shao 2018) and penalized eigen-decomposition algorithms. To establish the consistency of our method with a high-dimensional predictor and a multivariate response, we develop a new concentration inequality for sample MDDM around its population counterpart using theories for U-statistics, which may be of independent interest. Simulations and real data analysis demonstrate the favorable finite sample performance of the proposed method.

Data-driven slicing for dimension reduction in regressions: A likelihood-ratio approach

DIMENSION REDUCTION VIA ADAPTIVE SLICING

Sliced Regression for Dimension Reduction

On Sliced Inverse Regression with High-Dimensional Covariates

Slicing-free Inverse Regression in High-dimensional Sufficient Dimension Reduction

Flexible Dimension Reduction in Regression

Statistica Sinica Preprint No : SS-2016-0080 R 3 Title Flexible dimension reduction in regression

A Linear Discriminant Analysis model based on Sliced Inverse Regression lifting

Dimension Reduction in Regressions Through Weighted Variance Estimation

A Slicing‐Free Perspective to Sufficient Dimension Reduction: Selective Review and Recent Developments

Dimension Reduction in Regressions Through Cumulative Slicing Estimation

Dimension Reduction for Functional Regression with a Binary Response

Online Sufficient Dimension Reduction Through Sliced Inverse Regression.

A convex formulation for high-dimensional sparse sliced inverse regression

Projective Resampling Functional Sliced Inverse Regression

A Quasi-Residuals Method in Sliced Inverse Regression

On Spline Approximation of Sliced Inverse Regression

Sliced Coordinate Analysis for Effective Dimension Reduction and Nonlinear Extensions

On Consistency and Sparsity for Sliced Inverse Regression in High Dimensions

Slice Weighted Average Regression

On cumulative slicing estimation for high dimensional data