If you are attending #ICLR2026, please come and say hi to our group members presenting the papers 👇

We are looking forward to innovating with this massive compute support from @fz-juelich.de 🎉

Thank you! Great to be a part of relAI!

🥳Happy to share that we have three papers accepted to #ICLR2026. Congrats to our authors and see you in Rio🌴🇧🇷. Check the thread for highlights👇

Principal Investigators (f/m/x) for (i) AI in Biomedicine and (ii) AI Foundations

🚨 PI positions with an excellent package at Helmholtz Munich (deadline January 11). Please apply, please share! jobs.helmholtz-muenchen.de/jobposting/8...

We are celebrating our #NeurIPS2025 and #EurIPS2025 contributions below 🎊

We are looking for PhD students via the @ellis.eu and @munichcenterml.bsky.social PhD programs!

What if your CBM actually doesn’t look at attributes when it explains the classification decision? We suggest to use this dataset and method to generate realistic images with unusual attributes to stress-test explainability methods. @jessica-bader.bsky.social is at ICCV to talk to you about it!

Are you at #ICCV2025? 🎉 @iccv.bsky.social
EML Munich is presenting two poster papers—come say hi to our authors!
Details in the thread 👇

We got two papers of two orthogonal methods for interpretability and one paper of a GenAI method at @neuripsconf.bsky.social 2025 🔥 Please checkout the details in the thread below.

Thank you!

Post-hoc Probabilistic Vision-Language Models Vision-language models (VLMs), such as CLIP and SigLIP, have found remarkable success in classification, retrieval, and generative tasks. For this, VLMs deterministically map images and text descripti...

Unfortunately, our submission to #NeurIPS didn’t go through with (5,4,4,3). But because I think it’s an excellent paper, I decided to share it anyway.

We show how to efficiently apply Bayesian learning in VLMs, improve calibration, and do active learning. Cool stuff!

📝 arxiv.org/abs/2412.06014

I am hiring PhD students and/or Postdocs, to work on the theory of explainable machine learning. Please apply through Ellis or IMPRS, deadlines end october/mid november. In particular: Women, where are you? Our community needs you!!!

imprs.is.mpg.de/application
ellis.eu/news/ellis-p...

I am very sorry to hear this 😓

Thank you!

Thank you!

It is an honor to receive the ZukunftsWissen award from the German National Academy of the Sciences @leopoldina.org and Commerzbank-Stiftung 😊🙏🎉

Yes, you are right, we don’t know the official numbers yet. I was assuming that it is already 25%, but perhaps it is higher still.

Yes, I think so too. But I don’t think this is a good idea. 25-30% is already a rather low acceptance rate.

I am not sure. With the increasing number of submissions, to keep the number of accepted papers the same (due to space constraints), the acceptance rate needs to drop. This is how I read the email.

There are so many alternative solutions: half a poster space, an alternative official location, remote presentation with physical presentation in an alternative location, just a remote presentation etc.

I totally agree!

Due to physical resource constraints, we currently estimate that around 300–400 of the candidate papers recommended for acceptance by the ACs will need to be rejected. We seek the support of our 41 SACs in addressing this distributed optimization problem in a fair and professional manner.

NeurIPS has decided to do what ICLR did: As a SAC I received the message 👇 This is wrong! If the review process cannot handle so many papers, the conference needs yo split instead of arbitrarily rejecting 400 papers.

Congratulations Karsten @confusezius.bsky.social for a wonderful PhD journey that ended with a summa cum laude 🎉 It is rare that a PhD researcher creates a paradigm shift in a research group: I am grateful for everything that I learned from you 😉 Best of luck at Google DM in Zurich!

I love the #eurips initiative! But to live up to its potential, it should be accepted an official #neurips conference location (similar to Mexico City) and not just an addon! Then we would save C02, rather than adding to it!!! What can we do to achieve this?

🚨 Poster at #ICML2025!
How can LLMs really keep up with the world?

Come by E-2405 on July 15th (4:30–7:00pm) to check out WikiBigEdit – our new benchmark to test lifelong knowledge editing in LLMs at scale.

🔗 Real-world updates
📈 500k+ QA edits
🧠 Editing vs. RAG vs. CL

How can we circumvent data scarcity in the time series domain?

We propose to leverage pretrained ViTs (e.g., CLIP, DINOv2) for time series classification and outperform time series foundation models (TSFMs).

📄 Preprint: arxiv.org/abs/2506.08641
💻 Code: github.com/ExplainableM...

Thank you Julia! 👏

It is a great honor to receive the ZukunftsWissen Prize 2025 from the German Academy of the Sciences @leopoldina.org with generous support of the commerzbank foundation 🎉 This achievement wouldn’t have been possible without my wonderful group @eml-munich.bsky.social @tum.de @helmholtz.de