Welcome to Part 0!

Before we get started with creating super-powerful AI for beating challenges like Self-Driving Cars, Doom and Breakout, we need to first understand on an intuitive level the key concepts that go into reinforcement learning. And that's why we have included this Part 0 to make sure we're all on the same page.

The thing is we already know how we, humans, learn. We understand the concepts of intrinsic and extrinsic rewards that guide us to become better at things. For example, if you're playing bowling - you know that you need to hit the pins and a strike is a perfect shot. We know this because we are rewarded with points for hitting the pins down and we are "punished" with no points when your ball ends in the ditch. So your brain projects those conditions of the environment onto your actions and that's how you know when you are doing good and when - not so much. And that's how we learn.

But how do we explain that to an AI?

That's exactly what we will be covering off in this Part 0.

This part is intuition-only and the coding will come in later sections. So for now grab a cup of tea or coffee and enjoy!

See you inside!

Kirill & Hadelin