Content

Reinforcement Learning is one of the three main learning principles in machine learning and one of the most active research areas in artificial intelligence. It is a computational approach to learning in which an agent tries to maximize the total amount of reward it receives while interacting with a complex, uncertain environment.
This Bachelor seminar is based on the 2nd edition of the famous, seminal book on reinforcement learning written by Sutton and Barto (http://incompleteideas.net/book/the-book.html). The book introduces core topics of reinforcement learning from an artificial intelligence or engineering perspective, considering idealized learning situations, and evaluating the effectiveness of different learning methods. To effectively solve learning problems that are of scientific or economic interest, algorithms for machines are explored and evaluated through mathematical analysis and computational experiments. Compared to unsupervised or supervised learning approaches, reinforcement learning is more focused on goal-directed learning from interaction with the environment. The first part of the book addresses core concepts of reinforcement learning for problems with small state and action spaces, allowing for exact solutions using table-based methods. In the second part of the book these approaches are then extended using approximate methods for larger and more complex problems.

Learning Outcomes:

• Knowledge on different reinforcement learning algorithms
• Explain the underlying mathematical problem formulations and the implementation of the algorithms to solve them
• Gain insight into how to frame learning problems in the reinforcement learning framework
• Discuss practical applications of the theoretical frameworks
• Present the algorithms and mathematical problem formulations to an audience

Teaching form

In the seminar sessions students will present chapters of the book “Reinforcement Learning”, followed by discussions on the chapter topics.

Exam

Oral presentation and active participation

Enrollment

From 01/02/2026 to 01/03/2026, interested students can apply for a place in the seminar. The allocation of places will take place via the following central Moodle course of the faculty: https://moodle.ruhr-uni-bochum.de/course/view.php?id=62179. The final allocation of places will take place by 16/03/2026 at the latest. Complete your binding registration by registering for the seminar via FlexNow. Please note that following the above steps is mandatory. Enrolment via FlexNow without prior registration via the central Moodle course is not permitted.

Prof. Dr. Robert Schmidt Lecturer

(+49) 234-32-27300 robert.schmidt@rub.de NB 3/68

Course type

Seminars

Credits

3

Term

Summer Term 2026

Seminar

Takes place every week on Thursday from 08:30 to 10:00 in room IC 03/441.
First appointment is on 16.04.2026
Last appointment is on 23.07.2026

Knowledge of calculus, linear algebra, and probability concepts. Background in artificial intelligence, e. g. via the course “Introduction to Artificial Intelligence”.

Literature

Sutton, R. S., & Barto, A. G. (2018). Reinforcement learning: An introduction. MIT press.
Free online version: http://incompleteideas.net/book/the-book.html