Stagewise Newton, differential dynamic programming, and neighboring optimum control for neural-network learning | IEEE Conference Publication | IEEE Xplore