Category 4 - model-free

In RL, a fundamental distinction is if it is model-based or model-free. In model-free, we do not explicitly model the environment, or we do not know the entire dynamics of a complete environment. Instead, we just go directly to the policy or value function to gain the experience and figure out how the policy affects the reward:

• Policy and/or value function
  • No model