Policy-based Methods: RL algorithms that directly learn a parameterized policy and update it based on the rewards.