Model selection for inverse problems: Best choice of basis functions and model order selection

摘要: A complete solution for an inverse problem needs five main steps: choice of basis functions discretization, determination the order model, estimation hyperparameters, solution, and finally, characterization proposed solution. Many works have been done three last steps. The first two neglected a while, in part due to complexity problem. However, many problems, particularly when number data is very low, good selection become primary. In this paper, we propose within Bayesian framework. Then, apply method elastic electron scattering