Nonlinear regularization of inverse problems for linear homogeneous transforms by stabilized hard thresholding (Q2313805)

In this paper, the author considers a model of data with noise of the form \[ x_i = (Kf)_i + \varepsilon_i, \qquad i=1, \dots, 2^J \] where \((Kf)_i\) is the signal and the error terms \(\varepsilon_i\) satisfy the standard conditions of normality and independence. The estimation is based on the vaguelette-wavelet decomposition. The consistency and the asymptotic normality of the mean-square risk estimate are proved.