Convergent reinforcement learning control with neural networks and continuous action search | IEEE Conference Publication | IEEE Xplore