The main steps of a reinforcement learning model

The actions of the agent are the decisions that it can make. This is a limited set of decisions. As you will understand, the agent is just a piece of code, so all its decisions will need to be programmed controls of its own behavior.

If we think of it as a computer game, then you understand that the actions that you as a player can execute are limited by the buttons that you can press on your game console. All of the combinations together still allow for a very wide range of options, but they are limited in some way.

The same is true for our human baby learning to walk. They only have control over their own body, so they would not be able to execute any actions beyond this. This gives a huge number of things that can be done by humans, but still, it is a fixed set of actions.

Making the decisions

Now, as your reinforcement agent is receiving information about its environment (the state), it will need to convert this information...