Learning the structure of Factored Markov Decision Processes in reinforcement learning problems