Marcel Müller

I really liked that in an international context I can write to a professor with "Hi Dave". That gave me at least the illusion of a small hierarchy there. On the other hand, they give PhD students in Germany lot of responsibility as engineers. So that's fine too. Please don't distribute bad habits.

Wenn es nicht alle sind, würde ich tippen 12, 13 und 24.

Thank you very much for the comprehensive answer!!

Any tips you can share for good code documentation?

Thanks! Hopefully around October.

YouTube video by Marcel Müller Multi-Agent Reinforcement Learning for Deadlock Handling among Autonomous Mobile Robots

I don’t know if it’s groundbreaking (as claimed in the podcast), but I just submitted my PhD thesis.
If you're as impatient as I am for the defense, there's already a pretty cool podcast episode about it, generated with NotebookLM:
🎧 www.youtube.com/watch?v=YlpD...

Is this a bad joke or a seek for attention? If not, please search help immediately. There is nothing more valuable than life.

Thanks for sharing!

Thanks for sharing!

🤔 or maybe it's time to try RL on cryptocurrencies.

I think Cs:go is not too different from cs2. So I don't see the problem here. I reviewed also some paper with it a few days ago. I think it is still an interesting 3D environment with many problems combined. Is there actually some API for RL with cs2 like psyc2 for StarCraft2?

What is the composition of your classes like? In our programs, the representation of society already ends in the technical courses. For example, we started a new course "AI Engineering". Result: 25 % women, although we really tried to adapt the marketing to attract more women.

Das heißt dann wohl, dass YouTube die bessere KI hat. ;)

Cool paper. Which author are you? I saw some earlier paper from this research group, which was interesting.

Das ist ja übel. Bei YouTube gibt's kein Limit, oder? Habe da sicher 2000 h oder so.

Is there any paper feedback before 9th May?

Just don't get famous and you can handle it. ;)

The hype will pass. I am happy how people can express now their ideas. Endless funny memes. Studio Ghibli got also some free marketing here. I am curious if Google will accomplish to reach the broad mass with their very good models as well. Gemini was faster but not so popular with this feature.

Very interesting! Addresses directly my research. I will definitely read it. Did you also experiment with MARL for MAPF?

Ok, thanks. I will definitely join this discord.

Oh no, totally missed this competition. How can I avoid to miss this? Usually I checked AIcrowd. Any further MARL challenges in the future?

Da hört es ja leider nicht auf. Durch das Einscannen sind die PDFs so groß. Das war wieder klassisch deutsche Bürokratie: ausdrucken, unterschreiben, einscannen. Dabei haben wir ja digitale Signaturen...

Somewhat wrong answers from Claude etc. could be some interesting exam questions for students to find the errors and discuss.

Oof, what should the perfect president then look like?

Thanks for the clarification!

I want to submit the first time to RLC but don't get really the process yet. We submit an abstract and get a super fast response to know if we should submit the paper or how does it work?

Thanks, that's good to know and puts my mind at rest.

7 top papers a year is #slowscience? 🥲

Das "lustige" ist hierbei die Linie der IT, dass wir die Nexcloud nicht zur Dateiablage, sondern vor allem zur Kollaboration nutzen sollen. Also genau jener verkorkste Browsermodus.

What is happening over there in Tübingen? Why are so many ML people there?