RL is deeply entwined with several mathematical and dynamic programming concepts that could fill a textbook, and indeed there are several. For our purposes, however, we just need to understand the key concepts in order to build our DRL agents. Therefore, we will choose not to get too burdened with the math, but there are a few key concepts that you will need to understand to be successful. If you covered the math in the Chapter 1, Deep Learning for Games, this section will be a breeze. For those that didn't, just take your time, but you can't miss this one.

In order to understand the Q-Learning model, which is a form of RL, we need to go back to the basics. In the next section, we talk about the importance of the Markov decision process and the Bellman equation.