Sample-Efficient Reinforcement Learning: Exploration, Imitation, and Online Learning