Applying Policy Iteration for Training Recurrent Neural Networks

I. Szita, A. Lorincz

Recurrent neural networks are often used for learning time-series data. Based on a few assumptions we model this learning task as a minimization problem of a nonlinear least-square…