QuizMe

* השאלה נוספה בתאריך: 28-02-2025

		a. The training data needs to contain at least some trajectories that were generated using a good-enough policy
		b. The exploration/exploitation policy needs to be permissive enough to enable (in the limit) the evaluation of every action at every state
		c. We have to apply techniques such as importance sampling
		d. We must use a discount factor, where its values are 0 < γ < 1

For Q-learning to converge, which of the following options need to take place?