Robust Actor–Critic Learning for Continuous-Time Nonlinear Systems With Unmodeled Dynamics | Synapse