In the context of linear inverse problems, we propose and study a general iterative regularization method allowing to consider large classes of regularizers and data-fit terms. The algorithm we propose is based on a primal-dual diagonal {descent} method. Our analysis establishes convergence as well as stability results. Theoretical findings are complemented with numerical experiments showing state of the art performances.
@article{arxiv.1610.02170,
title = {Iterative regularization via dual diagonal descent},
author = {Guillaume Garrigos and Lorenzo Rosasco and Silvia Villa},
journal= {arXiv preprint arXiv:1610.02170},
year = {2017}
}
Comments
41 pages, 13 figures. 4-pages version of the paper available at http://opt-ml.org/papers/OPT2016_paper_19.pdf