Nonlinear estimation for linear inverse problems with error in the operator
The Annals of Statistics · 2008 · Volume 36 · Issue 1
DOI: 10.1214/009053607000000721
Abstract
We study two nonlinear methods for statistical linear inverse problems when the operator is not known. The two constructions combine Galerkin regularization and wavelet thresholding. Their performances depend on the underlying structure of the operator, quantified by an index of sparsity. We prove their rate-optimality and adaptivity properties over Besov classes.
Keywords
62G07, 65J20, degree of ill-posedness, Galerkin projection method, matrix compression, Minimax rate, Statistical inverse problem, wavelet thresholding
Enable JavaScript for citation tools and additional paper interactions.