Explore the third lecture in the TRIAD Distinguished Lecture Series featuring Sara van de Geer from ETH Zurich, focusing on the Debiased Lasso. Delve into the application of sparsity for establishing confidence intervals for parameters of interest. Examine the technique of using penalized estimators as initial estimators in a one-step Newton-Raphson procedure. Discover how functionals of this new estimator can be shown to be asymptotically normally distributed under certain conditions. Learn about the advantages of sparsity conditions in high-dimensional cases, particularly when the inverse Hessian of the problem is not sparse. This 58-minute lecture, presented on September 6, 2018, offers valuable insights into advanced statistical methods for researchers and practitioners in the field.
Overview
Syllabus
TRIAD Distinguished Lecture Series| Sara van de Geer | ETH Zurich | Lecture 3 (of 3)
Taught by
Georgia Tech Research