Hands-On Reinforcement Learning for Games

Constructing DQN

You can probably find a version of a DQN developed in every DL framework. The algorithm itself is an incredible achievement in learning, since it allows us now to learn continuous or infinite spaces/the infinite MDP. Open Chapter_6_DQN.py, and follow the next exercise to build the DQN sample:

The source for this example was originally derived from this GitHub repository: https://github.com/higgsfield/RL-Adventure/blob/master/1.dqn.ipynb. It has been modified significantly in order to match the previous samples in this book.

1. At this time, the samples have become too large to list in a single listing. Instead, we will go through section by section as we normally would. As usual, it is helpful if you follow along with the code ...