Module 3 is divided into three lectures (DS6-DS8). These lectures cover Chapter 8 - Chapter 13 but in a different order.

As before, the lectures jumps between chapters, so we suggest you look at the study instructions for each lecture to see what parts of the book are relevant for that lecture.

In this module, you will learn about function approximation for RL, policy gradient methods and model-based RL together with planning. The function approximation allows us to extend the algorithms we have covered so far to problems with continuous state and action spaces as well as provide possibly efficient methods for large discrete spaces. Policy gradient methods look at the control problem from a different (one may say more direct ) perspective than the methods covered so far. Model-based RL together with planning provides a 'modelling' perspective to the control problem.

In addition, you are encouraged to complement the study instructions by playing with the newly acquired concepts in Tinkering Notebooks of this module:

As described in Introduction to the Course, you will work mostly individually at your own pace, and in Study Groups. Take advantage of the study group to put forward your doubts, discuss the material, and learn more with your peers!

Each group is expected to bring their own reflections to the class session to take place at the end Module 3 (see the calendar).

In case you have generic doubts about Module 3, or if you find typos in the Study Instructions and /or Tinkering Notebooks, please use this Discussion Post to report them.

Contact teacher:

The contact teacher for Module 3 is Ayca Ozcelikkale - you can find the contact information of all teachers in People.