Information complexity-based regularization parameter selection for solution of ill conditioned inverse problems

We propose an information complexity-based regularization parameter selection method for solution of ill conditioned inverse problems. The regularization parameter is selected to be the minimizer of the Kullback-Leibler (KL) distance between the unknown data-generating distribution and the fitted distribution. The KL distance is approximated by an information complexity criterion developed by Bozdogan. The method is not limited to the white Gaussian noise case. It can be extended to correlated and non-Gaussian noise. It can also account for possible model misspecification. We demonstrate the performance of the proposed method on a test problem from Hansen's regularization tools.