Traditionally, methods for solving Sequential Decision Processes (SDPs) have not worked well with those that feature sparse feedback. Both planning and reinforcement learning, methods for solving SDPs, have trouble with ...