Model-Based Control

Once you do policy evaluation, you can do control.

You have two methods to improve your policy:

• Value Iteration
• Policy Iteration

#todo This is in Lecture 5: Model-Free Control, but I don’t remember seeing this, so you need to revisit Lecture 3.

Greedy Policy Improvement over V(s) requires model of MDP.

${\pi }^{\prime }(s)={\mathrm{argmax}}_{a\in \mathcal{A}}{\mathcal{R}}_s^a+{\mathcal{P}}_{s{s}^{\prime }}^{a}V(s^{\prime })$