Policy Gradient

تدرج السياسة

Definition

Reinforcement learning methods that improve the policy directly by ascending the gradient of the expected return with respect to its parameters.

Definition

Policy gradient methods optimise a parameterised policy directly by ascending the gradient of expected return, as in REINFORCE, actor-critic and PPO.

Mathematical equations

\nabla_\theta J(\theta)=\mathbb{E}_{\pi_\theta}\left[\nabla_\theta\log\pi_\theta(a\mid s)\,G_t\right]

Learn and review

Suggested terms to learn next