#goal oriented #reward maximization #policy gradient #learning curve #objective function #bellman equation