Policy Extraction From Q

Problem

Extract a greedy policy from an action-value table.

For each state row return the index of the largest Q value, breaking ties by choosing the smallest action index. An empty Q table returns an empty list.

Hint

Use a strict greater-than comparison so the earliest maximum wins ties.

Test cases

case 1

in: [[[1,3,2],[5,5,1]]]

expected: [1,0]

case 2

in: [[[0,-1],[-2,-3]]]

expected: [0,0]

case 3

in: [[[7]]]

expected: [0]

case 4

in: [[]]

expected: []

Starter codePython

def policy_from_q(Q):
    # Your code here
    pass

4 test cases · run your code against them in the editor