Question

i. Epsilon-Greedy is a technique that is used

a. to improve the model free Monte Carlo algorithms

b. to tune Q-learning algorithms to enable exploitation from the very beginning

c. to tune Q-learning algorithms to enable exploration all the time

d. to tune Q-learning algorithms to balance exploration and exploitation

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~`

ii. To obtain high Q-value as well as high average utility in Epsilon-Greedy technique

a. the best policy would be set Epsilon to zero (0) to enable exploration

b. the best policy would be to set Epsilon to one (1) to enable exploitation

c. the best policy would be set Epsilon to one (1) in the beginning to enable exploration, zero (0) at the end to enable exploitation and some value in between to combine exploration and exploitation

d. the best policy would be set Epsilon to a constant value like 0.5

Answer 1

Answer 1-

The correct Answer is —  to tune Q-learning algorithms to balance exploration and exploitation (option D)

Explanation:

• It is a simplex approach to equilibrate exploration and exploitation by selecting between exploration and exploitation at random.

Answer 2-

The correct Answer- the best policy would be set Epsilon to one (1) in the beginning to enable exploration, zero (0) at the end to enable exploitation and some value in between to combine exploration and exploitation(option C)

Explanation:

• It is fixed Epsilon to one (1) of the opening to modify exploration and zero (0) at the extremity to modify exploitation.

Note- Please do upvote, if any problem then comment in box sure I will help.