Friday
Monthly · Online via Teams · Free
Reinforcement Learning Coffee
An open, low-barrier meetup for anyone curious about reinforcement learning and sequential decision-making. Bring an early idea, a stubborn problem, a project, or a recent paper to explore together.
- When
- First Friday of every month
- Where
- Online via Microsoft Teams
- Who
- Students to senior researchers
Next session
- –sessions held
- –speaker affiliations
- 2024running since March
- 0 €free and open to all
Schedule & archive
Every talk so far
Pulled live from the organizers' schedule. Slides, papers and code are linked where speakers shared them.
About
Half-baked ideas welcome
The Reinforcement Learning Coffee meets on the first Friday of every month. The entry level is intentionally low and everyone is welcome, from newcomers to experienced practitioners. A session can be an initial idea or open problem, a more involved project, your latest paper or a paper idea, or simply working through a recent paper together.
The format grew out of the Machine Learning Coffee at CERN, where machine learning and reinforcement learning were introduced to the CERN accelerator complex. Talks since then have ranged from quantum optimal control and laser alignment to heat pumps, air traffic and chemical oscillators.
-
Low barrier
No polished results needed. Open questions and failed experiments make good sessions.
-
Real systems
A soft spot for RL on physical systems: accelerators, lasers, mechatronics, processes.
-
Across institutions
Speakers from universities, research institutes and industry, in Salzburg and remote.
Format
One hour, shaped by who shows up
Every session runs 60 minutes on Teams. The host picks one of three formats at the start, depending on how many people are in the call.
-
2 to 5 people
Round table
- 0–10 Short round: name, what you work on, one open question
- 10–30 Speaker walks through the work, slides optional, interrupt anytime
- 30–55 Open discussion, everyone on camera if possible
- 55–60 Next session and who presents
-
6 to 15 people
Talk and discussion
- 0–5 Welcome, news from the community
- 5–35 Talk, about 25 minutes, clarifying questions in between
- 35–55 Discussion led by the host, raise hand or unmute
- 55–60 Next session and call for talks
-
16 people or more
Seminar
- 0–5 Welcome, microphones off by default
- 5–40 Talk, about 30 minutes, questions go into the chat
- 40–55 Host reads out the chat questions, then raised hands
- 55–60 Next session; slides are linked in the archive if the speaker agrees
Recordings only with the speaker's consent. Early ideas and unfinished work stay in the room.
Stay in the loop
Get a reminder before every session
Join the mailing list for the monthly reminder, the talk details and any last-minute changes.
Present something
Have an idea, a result, or a question?
Every stage is welcome, from a first sketch to a finished preprint. Write to sarl.ida.plus@gmail.com to pick a Friday.
Organized by
SARL at the IDA Lab
The Smart Analytics & Reinforcement Learning team at the IDA Lab, Paris Lodron University of Salzburg (PLUS).