Exercise 5: Q-learning
Paper Exercises
In this exercise we will discuss key concepts such as
- The Bellman equation
- Q-learning
- Epsilon-greedy exploration
- Optimal Q-values, Q*
- Learning rate and discount factor
Paper exercises and solutions can be found in the exercise compendium.
It is recommended that you complete this paper exercise before moving on to the computer exercise.
Computer Exercises
Can be found directly here;
Q_Learning_Warehouse.ipynb Download Q_Learning_Warehouse.ipynb
Or on the remote lab computers like previous computer exercises, in the lab called frtn75-rl.
The exercise explores:
- Warehouse problem (Q-learning maze)
Solution is available at Q_Learning_Solution.ipynb Download Q_Learning_Solution.ipynb or on the jupyterhub server.