QUESTION $1$

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.

True

False

QUESTION $2$

When an agent remains within the same environment region for some time it will have similar experiences. This can bias the learning algorithm towards that region, and it will not perform well outside that region. In order to overcome this problem instead of using the most recent learning experiences the agent learns based on a "replay buffer" holding only its very distant past experiences

True

False

QUESTION $3$

In order to measure the performance of a Reinforcement Learning agent we sum up the rewards it is getting.

True

False

QUESTION $4$

The "credit assignment problem" refers to the problematic fact that a Reinforcement Learning agent has not direct way of knowing which of its previous actions are contributing to a given reward.

True

False

QUESTION $5$

Regarding the Discount Factor:

The Discount Factor can be thought as a measure of the value we give to the future relative to the present.

De Discount Factor greatly alfects the optimal policy.

All of the above are true.