Special Topics in Deep Reinforcement Learning

Dear students, 

We would like to conclude the seminar with a final meeting, during which we will present the results of your projects. 

The meeting will take place on Monday, October 21, at 4.15 pm in the DFKI building D3.1 (Altbau) room Leibniz (+1.29). If you don't know where the room is, meet us at 4.10 pm in the lobby. 

 

Looking forward to see you on Monday, 

the STDRL seminar team

Dear students, 

 

We will have news regarding your projects shortly. 

This time, we would like to ask for your help. We are currently conducting a user study about a tool we developed and, therefore, urgently need users. 

The study is running until Thursday (Oct 24) next week, and you can participate every day from 9 am to 5 pm (lunch break from 12 - 1) in DFKI main building (D3.2) room Bledsoe (+2.07). We have enough computers for several participants to do the study simultaneously. If you find time to participate sometime this week, this would be very helpful. Besides our everlasting gratitude, you will be compensated with 20 €. You can either: 

- drop by the room Bledsoe between 9 am and 5 pm (except lunch 12 - 1)
- write me an e-mail to arrange a time (timo_philipp.gros@dfki.de)
- send me a message on Teams to arrange a time. 

 

Thank you in advance, 

Timo 

Dear students, 

We have realized that none of the groups have made major progress on the project so far. As we already said at the seminar kickoff, you are free to decide when you work. However, the fact that none of the groups have made significant progress worries us. 

If you need any help or assistance, don't hesitate to contact Nicola or Timo. We are happy to discuss any project-related questions. 

 

We are looking forward to seeing your results!

 

Best,

the STDRL seminar team

Dear students, 

we have published the grades of the seminar talks, they are visible on your personal status page. 
Please recall, that the final seminar grade does not only consist of the talk grade. So make in particular sure to work also on the practical project. 

Best regards, 
The Special Topics in DRL - Team

Dear Students,

Please register in the MoSi Gitlab using your university e-mail under the following link: https://mgit.cs.uni-saarland.de/users/sign_in

Afterward, one person in your group must send an e-mail to Nicola containing the user names of all group members, such that he can create your repositories.

Best,

the STDRL seminar team

Dear Students,

We would like to emphasize two points about the practical project that may not have been clear during the presentation:
1.) You are not allowed to use libraries with pre-implemented DRL algorithms; you must implement them yourselves. All required libraries will be contained in the provided virtual environment.
2.) Please ensure that evaluating your submitted agents is reasonably fast. We cannot accept submissions that require us to evaluate agents for days. 

Best,
the STDRL team

Dear students, 

Please remember that you must also register for this seminar in the LSF. The deadline for this registration is May 14. (TODAY!) If you can't register in the LSF, please contact us immediately. 

Additionally, please remember to send a preliminary version of your slides to your supervisor. As the first session takes place next Tuesday (May 21), this deadline is also today (edit: for those students who give their talk next week). 

 

We are looking forward to your talks. 

 

Best, 
the STDRL seminar team

Dear students, 

 

Please remember that you must also register for this seminar in the LSF. The deadline for this registration is May 14. (It's already next week!)

Please do not forget this additional registration. If you forget, we cannot let you participate in the seminar any further, even if we'd like to.  

 

Best, 
the STDRL seminar team

 

Dear students, 

you can now find the slides of yesterday's kick-off under Materials. 

Due to a scheduling conflict, we will need to swap the order of Causal Reinforcement Learning and Safe Reinforcement Learning: in contrast to the numbers on the slides, the talks of Causal Reinforcement Learning will be scheduled first. To avoid confusions, we will leave the numbers of the topics unchanged (see also @ Topics & References).

Sorry for the amendment. We will release an exact schedule as soon as we assigned the topics. If anything is unclear, don't hesitate to contact us. 

 

Best, 

the STDRL seminar team