Return to Article Details Accelerating Reinforcement Learning by Composing Solutions of Automatically Identified Subtasks Download Download PDF