We study the effect of using weaker forms of data-fidelity terms in generalized Tikhonov regularization accounting for model uncertainties. We show that relaxed data-consistency conditions can be beneficial for integrating available prior knowledge.
翻译:暂无翻译