The Art of Reinforcement Learning: Fundamentals, Mathematics, and Implementations with Python

© The Author(s), under exclusive license to APress Media, LLC, part of Springer Nature 2023

M. HuThe Art of Reinforcement Learninghttps://doi.org/10.1007/978-1-4842-9606-6_7

7. Nonlinear Value Function Approximation

Michael Hu¹  

(1)

Shanghai, Shanghai, China

 

In this chapter, we will explore the use of nonlinear methods to approximate value functions in reinforcement learning. Value functions play a crucial role in reinforcement learning as they estimate the expected reward an agent can obtain in a given state and follow the policy afterwards. In the previous chapter, we discussed linear methods that rely on constructing a feature vector and computing a weighted combination of features as the state or state-action value. However, these methods have ...