B0315
Title: Fourier transform sparse inverse regression estimators for sufficient variable selection
Authors: Jiaying Weng - Bentley University (United States) [presenting]
Abstract: Sufficient dimension reduction aims to reduce the dimension of predictors while maintaining the regression information. Recently, researchers study an impressive range of sparse inverse regression estimators. Nonetheless, conspicuously less attention has been given to the multivariate response with high-dimensional covariates settings. To fill the gap, we investigate Fourier transform inverse regression approach via regularized quadratic discrepancy functions. Theoretically, we establish the consistency and oracle property for the proposed estimators. We propose an iterated alternating direction method of multipliers (ADMM) algorithm to estimate two target parameters simultaneously. We derive the explicit solution for each step of the ADMM algorithm. Numerical studies and real data analysis confirm the theoretical properties and yield superior performance of our proposed methods. In specific, our proposal has higher support recovery rates compared to the state-of-the-art approach.
