FS2023: 62514 Seminar Advanced Topics in Reinforcement Learning and Decision Making

The course starts with a few introductory lectures, before moving on to some advanced concepts. These will be synchronised with Tuesday's lectures in the basic course. After the introductory lectures are complete, we will have weekly readings of papers. All students will read the paper for the week, and submit paper reviews, summarising the main points, the weaknesses and strengths of each paper, as well as any open problems. The paper will then be presented by one of the students, and the remaining students will ask questions.

There is no fixed syllabus, but some possible topics include:
- Complexity of Markov decision processes
- Stochastic approximation and reinforcement learning
- Deep reinforcement learning and function approximation
- PAC and Regret bounds
- Multi-agent problems
- Fairness
- Privacy

Highly motivated students that will do a MSc thesis in the field, or that may later wish to do a PhD.

Previous exposure to probability, linear algebra and calculus is necessary.
If you have already taken a course in reinforcement learning, though, you can follow the seminar on its own.
if not, it is highly recommended to follow the course "Reinforcement Learning and Decision Making Under Uncertainty" in parallel. The paper sets to be discussed will be aligned with the lecture scheduled in the other course.

Deutsch

This work has all rights reserved by the owner.

Unbegrenzt – wenn online geschaltet

Sie können diesem Kurs direkt beitreten.

Unbegrenzt

Anmeldename
Vorname
Nachname
E-Mail