Title: Asymptotic normality and bias correction in high-dimensional statistical inference with regularized estimators Authors:  Jing Zhou - University of East Anglia (United Kingdom) [presenting]
Abstract: Bias correction stands as an important technique for high-dimensional statistical inference using regularized estimators. The idea of this technique is to correct the bias caused by the regularizer and demonstrate an asymptotic normality of a complete sparse parameter vector. This line of research is advantageous in that it considers selection uncertainty and allows for variability of the nonnull components of the parameter vector. This is especially attractive because the asymptotic oracle properties of the regularized estimators are unlikely to hold in finite samples. Obtaining the asymptotic normality of the biased corrected regularized estimators provides flexibility in both estimation and variable selection. The potential of hypothesis testing is showcased using the bias correction technique for the regularized M-estimators.