Inverse Reinforcement Learning through Structured Classification