Regarding differences between Reinforcement Learning and the usual Supervised or Unsupervised learning:

In Reinforcement Learning the goal is to find a good policy as opposed to unsupervised or supervised learning where the goal is usually to find patterns in the data.

In Reinforcement Learning the agent is not explicitly given the answer, instead it must find it by trial and error. Note that in supervised learning you are given the target answer.

While unsupervised learning has zero supervision, Reinforcement Learning uses an indirect form of supervision through the rewards, which tells whether progress is being made or not.

In Reinforcement Learning, unlike unsupervised or supervised learning, training instances are usually not independent especially while the agent remains in the same region of the environment where consecutive observations are highly correlated.

All of the above are true.

None of the above are true.