This chapter tried to cover a lot of material, but did not provide many practical new examples. Be sure to try some of the following exercises to build on your already growing experience as an RL guru:
- Return to the Banana Collectors and add several more agents using Ctrl + D or (Command + D on macOS). How can you keep adding agents without the training scenario lagging or slowing down too much?
- Convert the Soccor example to use Imitation Learning for one of the players. If you select the Goalie type, then set one Goalie player as a Teacher and one as a Student.
- Convert the GridWorld example to use Imitation Learning. Create a new Teacher agent and convert the existing agent into a Student. We covered this example in some detail, ...