Why is the actor-critic algorithm limited to using on-policy data? Or can we use the actor-critic algorithm with off-policy data?
1 Answer
$\begingroup$
$\endgroup$
1
It's because, in the actor-critic algorithm, the objective function is an expectation under the $\tau$ of the policy. If we want to use off-policy data, we have to resort to importance sampling relative to the other policy.