Policy Gradient Methods

Actor-Critic Methods

Hybrid approach combining an actor that learns the policy and a critic that estimates the value function, reducing the variance of policy gradient estimates.

← Tillbaka