An Improved Population-Based Incremental Learning Method for Objects Buried in Planar Layered Media

An evolutionary algorithm, the estimation of distribution algorithm (EDA), is used to reconstruct the objects that buried in planar layered media. It is essential that fast forward solvers be used to solve the forward scattering problem for the nonlinear inverse scattering methods, since it can avoid errors by approximation. The EDA is a predominant all-round optimizing method in the macroscopic simulation of evolution process species of nature. Recent studies have shown that the EDA provides better solution for nonlinear problems than the microscopic evolutionary algorithm, such as genetic algorithm (GA) in some cases. The EDA is simpler, both computationally and theoretically, than the GA. We discuss how this can be used to calculate the permittivity and conductivity of the targets. We show preliminary results indicating the potential of reconstruction for buried objects. Compared with other methods, the experiment result shows that the EDA algorithm reduces the number of iteration.