Regularization Methods and Inverse Problems: An Information Theory Standpoint