Which of the following statements are true?

i. In Finance, reinforcement learning might be an appropriate framework for future expectations, if this future is influenced by the actions of rational agents.

ii. Modelling corporate defaults is an exercise in Unsupervised Learning because we do not know the future.

iii. Reinforcement Learning is a suitable framework for portfolio optimization, even though it can also be done with Supervised Learning using some pre-specified models of the world.

iv. The difference between (direct) Reinforcement Learning and Inverse Reinforcement Learning is that in the latter case, there is no information about rewards received by the agent.