Our paper on non-convex optimization for end-to-end learning and control has been accepted at ICLR 2021!