RAML Meetup

Jan 15, 2026, 10:00 AM → Jan 16, 2026, 3:00 PM Europe/Berlin

JvF25/3-302 - Co-Working Space (Lamarr/RC Trust Dortmund)

Jian-Jia Chen, Raphael Fischer (TU Dortmund University), Sebastian Buschjäger (Lamarr Institute for ML and AI, TU Dortmund)

Thu, January 15

10:00 AM Arrival

Grab a coffee and some snacks until everyone is there :)

1 Welcome

An introduction round and guided tour through the next 1.5 days.

Speaker: Raphael Fischer (TU Dortmund University)

RAML Meetup Pictures.zip

raml_meetup_slidedeck.pdf

2 Area and Lamarr Essentials

Kicking things off, we exchange our thoughts on Resource-Aware ML, collect the most important conferences and deadlines, and learn about the bigger context of Lamarr!

Speakers: Raphael Fischer (TU Dortmund University), Stefan Michaelis (Lamarr Institute - TU Do)

REAML-Lamarr-Essentials-and-direction.pdf

• Resources are as important as the employment of machine learning pipelines themselves!
• We can do theory for everything, along the ML pipeline!
• When we talk about theory for resource awareness, do we talk about optimization theory? When wouldn’t we strive for optimization?
• Some of Lamarr’s RAML conferences/journals: EmSoft, MLsys, AutoML, TCAS, FAccT, SODA, EC, QCE, ECML, COLT
• Talking Points: Identify the main public debates on resource awareness, and discuss them with the PR department.

3 Overview of the Resource-Aware Machine Learning Area at Lamarr

Speaker: Jian-Jia Chen

12:00 PM Lunch Break

We will have lunch at the Cantina of JvF 25 (this is the same building on the ground floor)

4 Interactive Clustering

Yes, you read correctly - we will do clustering. High-dimensional, highly personalized with yet to be discovered distance metrics. In all seriousness: We hope to do some interactive clustering with post-its on the whiteboard that will (hopefully) lead to nice discussions on your research ideas and topics.

Speakers: Jian-Jia Chen, Raphael Fischer (TU Dortmund University)

Important Clusters:

• Probabilistic theory and modeling (Friday session) 
• Performance trade offs (Friday session)
• Optimization (Friday session)
• Time series
• Transformers and LLMs / model distillation
• Tabular data & AutoML (Eggensperger and Feurer Groups)
• Beyond worst case analysis.
  

 

2:40 PM Coffee Break

5 (Some) Highlight Presentations

Here the ReAML PIs will give overviews for the efforts of their group!

Speakers: Heiko Röglin, Jian-Jia Chen, Katharina Eggensperger (TU Dortmund), Matthias Feurer (TU Dortmund), Thomas Kesselheim

6 Wrap-Up

Speakers: Jian-Jia Chen, Raphael Fischer (TU Dortmund University)

See Wrap-Up Day 2!

4:15 PM Long Break before dinner

You can check into your hotel, check your E-mails, make sure your cat is fed at home etc.

7 Dinner

Restaurant Namu – Innovative Korean Cuisine, located in the Dortmund city center. Reserved for 16 people on "Fischer"

Fri, January 16

9:00 AM Arrival

You know the drill by now: Get some coffee and say hi to everyone arriving

8 2026 pitches

You were asked to prepare 1 slide on your research plans for 2026. Now its the time to get them out and present them!

Speakers: Jian-Jia Chen, Raphael Fischer (TU Dortmund University)

2026_pitches.pdf

10:30 AM Pre-Lunch Break

9 Clustered-Dating

You have put in a lot of work to cluster yourselves the day before, and now it's time to capitalize on it! Dont worry: This is about research only - we do not expect you to find a new life partner. Although it might be a nice success story for Lamarr....

Speakers: Jian-Jia Chen, Raphael Fischer (TU Dortmund University)

12:00 PM Lunch-Break

We will have lunch at the Cantina of JvF 25 (this is the same building on the ground floor)

10 Idea building

Now that you (hopefully) have found your research partner, its time to double-down. We ask you to form small groups with 2-N people in which you build a more concrete research idea. This can be a paper idea, a project proposal, or just some random idea that you want to quickly try out. Try to come up with a short pitch of
a) What you want to do
b) How you want to do it
c) What the timeline is
d) (optional) What resources you need to achieve it

Speakers: Jian-Jia Chen, Raphael Fischer (TU Dortmund University)

11 Idea pitches

Well, those ideas are not going to present themselves. So its time to pitch them to the group :-)

Speakers: Jian-Jia Chen, Raphael Fischer (TU Dortmund University)

Topics: 

• Sustainability
  • Sustainability means effective use of resources
  • Sustainability means to have a long lifetime span / reusability
  • We can improve sustainability from hardware level or software level
  • Relation to flexible and reusable models
  • For sustainability, people don’t always consider the entire software pipeline, and only focus on energy consumption, CO2 emissions, etc.
  • If you want to be sustainable you will have to compromise on something => Performance trade-offs!
  • Every model's energy consumption is just an approximation to the realty: Ground-Truthing AI Energy Consumption
  • Would you pay more money for being more sustainable?
  • Green AutoML => Can we quantify and tune for reusability / flexibility?
  • We should care about human, and look at them as resources => Build models to learn the decision making of users?
• Performance Trade-Offs
  • Various trade-offs exist: Energy, Time, Accuracy, Memory, Cost, Traffic, Quality, Explainability, Fairness, Security, Privacy, ...
  • Training vs Inference vs other tasks?
  • Meta Study on ML Trade-Offs across the field?
  • Once we have data on trade-offs, we can perform multi-objective optimization
    • Maybe we can organize a respective workshop on the topic?
• Optimization
  • ?
• Probabilistics
  • ?

12 Wrap-Up

Time to wrap-up the seminar. Thank you all for being here

Speakers: Jian-Jia Chen, Raphael Fischer (TU Dortmund University)

• Area Organization:
  • RAML - Lamarr Institute Mattermost Chat for central communication => PIs can reach out to create accounts for students (see Lamarr Institute - Wiki)
  • Central slide deck: lamarr-institute/raml-slides: Slidedeck for the Resource-Aware ML area
  • Mission & Example Approaches given in Lamarr Institute - Wiki
  • Follow-up on Area Focus Clusters and Ideas via Google Doc: RAML 2026 - Focus Areas and Ideas
• Meetings
  • PI & Postdoc bi-monthly area coordination meeting at 11:00
  • Monthly seminar at 11:30 → 30 min freestyle on last interesting paper
  • Last / first WEEKDAY (to be clarified) of each month
  • Central Area Meetups as part of Lab Visits & Co
• TODO
  • All PIs: Review the Lamarr RAML Overview Website and RAML Wiki entry - do you feel represented? What topics can be updated / removed?
  • Who is interested in identifying and answering Talking Points in RAML with the PR Team? => Get back to Raphael & JJ!
  • Next Area Meetup @ Lab Visits 2026.1 (23-24. März 2026)