CFE 2019: Start Registration
View Submission - CMStatistics
Title: Smoothed-residual stopping for statistical inverse problems via truncated SVD estimation Authors:  Bernhard Stankewitz - Humboldt University of Berlin (Germany) [presenting]
Abstract: The purpose is to examine under what circumstances adaptivity for truncated SVD estimation can be achieved by an early stopping rule based on the smoothed residuals $ \| A^{\alpha} ( Y - A \hat{\mu}^{(m)} ) \|^2 $. Lower and upper bounds for the risk are derived, which show that moderate smoothing of the residuals can be used to adapt over classes of signals with varying smoothness, while over-smoothing yields suboptimal convergence rates. The theoretical results are illustrated by Monte-Carlo simulations.