Nonlinear State Space Estimation With Neural Networks And The Em Algorithm

In this paper, we derive an EM algorithm for nonlinear state space models. We useit to estimate jointly the neural network weights, the model uncertainty and the noisein the data. In the E-step we apply a forward-backward Rauch-Tung-Striebel smootherto compute the network weights. For the M-step, we derive expressions to compute themodel uncertainty and the measurement noise. We find that the method is intrinsicallyvery powerful, simple, elegant and stable.iContents1 Introduction 12 ...