Reinforcement learning can be formalized in terms of ____ in which the agent initially only knows the set of possible _____ and the set of possible actions.
- Markov decision processes, objects
- Hidden states, objects
- Markov decision processes, states
- objects, states
