Machine Learning with Reinforcement for Optimal and Adaptive Estimation Problems in Navigation Applications

The statement and solution of the nonlinear estimation problem by the minimization of the root-mean-square criterion based on neural algorithms are given both within the Bayesian approach using a training set for supervised learning in off-line mode and within the framework of the least squares method in the presence of a measurement equation for the reinforcement learning on-line mode. An illustrative example is considered.