1: \begin{abstract}

2: This paper is concerned with a novel regularisation technique for solving linear

3: ill-posed operator equations in Hilbert spaces from data that is corrupted by

4: white noise. We combine convex penalty functionals with extreme-value

5: statistics of projections of the residuals on a given set of sub-spaces in the

6: image-space of the operator. We prove general consistency and convergence rate

7: results in the framework of Bregman-divergences which allows for a vast range of

8: penalty functionals.

9: 

10: Various examples that indicate the applicability of our approach will be

11: discussed.  We will illustrate in the context of signal

12: processing that the presented method constitutes a fully data-driven method for

13: denoising that additionally exhibits locally adaptive behaviour.

14: \end{abstract}

15: