We investigate the use of deep artificial neural networks to approximate the optimal state-feedback control of continuous time, deterministic, non-linear systems. The networks are trained in a ...