Summary

In this chapter, we covered three important RL application areas: Personalization, marketing, and finance. For personalization and marketing, this chapter went beyond the bandit applications that are commonly used in these areas and discussed the merits of multi-step RL. We also covered methods such as dueling bandit gradient descent, which is helpful to achieve conservative exploration to avoid excessive reward losses, and action embeddings, which is helpful to deal with large action spaces. We concluded the chapter with a discussion on finance applications of RL, its challenges, and introduced the TensorTrade library.

The next chapter is the last application chapter of the book, in which we will focus on smart city and cybersecurity. Take a break, and join us on the other side!