Frozen Lake Example in RL (Again)
Hello, everyone!
After spending some time with RL, I got a better understanding on Q-learning. Therefore, I have rewritten a solution to the GridWorld problem.
But difference from the previous article is that I did not waste time to generalize the program or create an animation.
Just plain (and more accurate) calculations only using numpy. Also it is more descriptive.
It is again in the form of Jupyter Notebook. In the last cell I have included a “history browser” to keep track of the changes made by the algorithm. I hope you will find it useful.
Oh, and one last thing: I am attaching a PDF file of empty Q table of the default Grid World example. So, if you want to solve it yourselves it will help a lot (it did to me, but I got bored in the half of the first iteration). You can use it as you like.
Resources