Prof. Dr. rer. nat. Marius Lindauer
In recent years, Reinforcement Learning (RL) has produced some of
the most impressive results in the realm of Machine Learning (ML), especially in game playing
(as with the game of Go) and robotics (e.g. RoboCup or autonomously navigating robots).
Its view of the ML model as an agent acting within an environment, allows for
learning by trial and error and therefore reasoning beyond human expert
RL is a quickly evolving field with new algorithms and applications being
Therefore this course will begin with teaching the mathematical foundations of RL
and give an overview over the field's development up until today.
At the end of the lecture you will be able to understand the current state of RL
research as well as reason about the theoretical foundations of different RL
Via the accompanying exercises you will be introduced to implementing several RL
algorithms as well as the general RL pipeline, including learning environments,
agent evaluation and hyperparameter settings.
At the end of the semester you will apply your new skills to an interesting RL project of your choice.
We strongly recommend that you know the foundations of
- Machine Learning
- Deep Learning
in order to attend the course. You should have attended at least one other course for ML and DL in the past.
Lecture topics include:
- Markov-Decision Processes
- Value-function Approximation
- Policy Search
- Model-based RL
- Deep RL
by Richard S. Sutton and Andrew G. Barto
This course as well as the exercises will be in English only.
The course will be held online with live sessions for both lecture and exercises. There will be opportunities for questions and discussion during the lecture, additionally there will be an online forum and the exercise sessions with extra examples for the current topics. Each week there will be a programming task to be completed at home in groups of up to three students. These tasks will also be discussed during the exercises. At the end of the semester, you will be free to choose a task to tackle using RL.