Hands-On Intelligent Agents with OpenAI Gym

Deep n-step advantage actor-critic algorithm

We observed that the actor-critic algorithm combines value-based methods and policy-based methods. The critic estimates the value function and the actor follows the policy, and we looked at how we can update the actor and the critic. From our previous experience in Chapter 6, Implementing an Intelligent Agent for optimal discrete control using Deep Q Learning, we naturally got the idea of using a neural network to approximate the value function and therefore the critic. We can also use a neural network to represent policy , in which case parameters are the weights of the neural network. Using deep ...