PG: Policy Gradient
摘要
The policy optimization algorithms in Chaps. 2–6 use the optimal value estimates to find the optimal policy, so those algorithms are called optimal value algorithm. However, estimating optimal values are not necessary for policy optimization.