Hands-On Artificial Intelligence for Beginners

Value functions

A value function helps us measure the expected reward at certain states; it represents the expected cumulative reward from following a certain policy at any given state. There are two types of value functions used in the field of reinforcement learning; state value functions V(s) and action value functions .

The state value function describes the value of a state when following a policy. It is the expected return that an agent will achieve when starting at state s under a policy π. This function will give us the expected reward for an agent given that it starts following a policy at states:

Let's break down what this function ...