Deep Reinforcement Learning (Master) [SS21/WS21]

Seminar Facilitator: Hendrik Annuth

Description: Reinforcement Learning has proven to be a useful tool, in certain areas of computation. In this seminar, we will dive into the technical aspects of this topic and analyze algorithms as well as recent success stories.

Language: The scientific literature and the scientific field of data science is vastly represented in English language. Therefore, the written version of the presentation and the presentation material has to be written in English language. However, the presentation itself can be done in German language, even so, English is preferred. The English skill-set of participants will not influence the evaluation of the lecture. Participants are expected to prepare their presentations, thus, if the presentation content is hard to understand, due to ill-prepared presenters, this will influence the evaluation of the lecture for obvious reasons, however. 
 
Expectations: Every participant is expected to do a 45min presentation followed by a discussion on their topic, and create a 10-page written version of their lecture. All presentations in the seminar are held under compulsory attendance (Anwesenheitspflicht).  Participants are expected to communicate with one another to avoid overlapping topics and discuss who will present which topic. The submission date for the written submissions one week before your presentation. The document needs to be conform with the PDF file standard. The written submission is a preparation for the master’s thesis, and is thus, expected to aim for scientific standard. Texts should be easy to understand and if necessary, properly illustrated. Participants are expected to name important sources and authors, and quote publications. Participants are expected to use source material that meets scientific standard, such as scientific literature and other forms of scientific publications. The written submission cannot exceed 10-page length. The evaluation of participants lectures is not only based on their expressed knowledgability in their given field, but also on their didactical preparation of their presentation and written material. Thus, participants are expected to have their explanations of content, representations, and illustrations ready and practiced. For the first 8 topics presentations are expected to contain the presentation of explicit algorithms including an example of their working. For the last 4 topics this is optional, however, they need to include a detailed description of the inner working of the combinations of different networks used. 

Source: As a starting point for most topics please consider the lecture from David Silver 

Which again is strongly based on the following books 

An Introduction to Reinforcement Learning; Sutton, Barto; MIT Press, 1998 

Algorithms for Reinforcement Learning; Szepesvar; Morgan and Claypool, 2010 

Nice introduction to RL from OpenAI


 

Available Topics: 
 
1. Dynamic Programming in Reinforcement Learning (Shiva Donthi)

2. Markov Decision Process (MDP) (Amar Bolkan)

3. Model-Free Prediction (Sarah Fränkel)

4. Reinforcement Learning based on Value Functions (Michele-Denise Friedlein)

5. Policy based Reinforcement Learning (Chalga Amanpreet)

6. Model based Reinforcement Learning (Fynn Kraft)

7. Exploration and Exploitation (Thorger Dittmann)

8. Reinforcement Learning in Games (Marco Schneider)

9. Reinforcement Learning in Robotics (Khalil Sander)

10. OpenAI Five 

11. Alpha Go (+Zero) (Pablo Rometsch)

12. AlphaStar (Vincent Vogt)


Distribution of Topics: 30.06.2021 12:30-13:45
Virtual Meeting Room 
 https://fh-wedel.zoom.us/j/99879013027?pwd=YW96QnhOWlhtQ2R5N2hqeGRXNE1HQT09