Reinforcementlearninganddynamicprogrammingusingfunctionapproximators