EcoSta 2018: Registration
View Submission - EcoSta2018
A0657
Title: On weighted inverse regression ensemble for sufficient dimension reduction and sufficient variable screening Authors:  Zhou Yu - East China Normal University (China) [presenting]
Abstract: Based on the conditional characteristic function of the response given the predictors, we introduce weighted inverse regression ensemble (WIRE) as a unified framework for dimension reduction and sufficient variable screening. Unlike classical sufficient dimension reduction estimators and existing sufficient variable selection procedures, WIRE is slicing-free and is readily applicable in the case of multivariate response. Under the setting with fixed predictor dimensionality, the $\sqrt{n}-$consistency of the sample level WIRE estimator is established for dimension reduction. We further propose a forward regression algorithm based on WIRE for ultra-high dimensional feature screening, which enjoys the model-free feature screening consistency when $p$ diverges at an exponential rate of $n$. The superior finite-sample performances of our proposals over existing methods are demonstrated through extensive simulation studies and the analysis of the Cancel Cell Line Encyclopedia data set.