Agents learn optimal behavior by interacting with an environment and maximizing reward.
No models in this category are strong on that dimension yet.