IRL Max-Entropy Gradient

Problem

Compute the max-entropy inverse RL gradient as the difference between expert and policy feature expectations: expert_feats[i] - policy_feats[i].

Hint

The gradient vanishes when the policy matches the expert's feature counts.

case 1

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

expected: [1,2]

case 2

in: [[1],[1]]

expected: [0]

case 3

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

expected: [-2,1]

Starter codePython

def maxent_irl_gradient(expert_feats, policy_feats):
    # Your code here
    pass

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