psteinb.bsky.social

psteinb.bsky.social

@psteinb.bsky.social

Using Machine Learning for Matter Research @helmholtzai.bsky.social

359 followers346 following56 posts

Deadline extended to May 6, 2026! Go and submit your experiences, please.

psteinb.bsky.social@psteinb.bsky.social

Don't forget to submit your poster idea on "Open challenges in SBI" at our workshop in Munich on June 8, 2026! 👉 events.hifis.net/event/3955/a... (deadline Apr 29, 2026) We are happy to have @deismic.bsky.social as the keynote speaker. Come join us and show us your #sbi work! @sbi-devs.bsky.social

psteinb.bsky.social reposted

TMLR's 2025 annual report! The word of the year is growth: whether it's our number of submissions, published papers, or reviewers, pretty much everything has increased ~50%. EXCEPT: our review period, which is still shorter than NeurICMLR (91 vs 126 days) Read on for more! 1/n

This is so cool! 👇👇👇

From hackathon to release: sbi v0.25 is here! 🎉 What happens when dozens of SBI researchers and practitioners collaborate for a week? New inference methods, new documentation, lots of new embedding networks, a bridge to pyro and a bridge between flow matching and score-based methods 🤯 1/7 🧵

psteinb.bsky.social reposted

Heute vor 6 Jahren wurde Walter #Lübcke von einem Rechtsradikalen ermordet. Walter Lübcke steht für einen Konservatismus, der heute so sehr fehlt. Ein Konservatismus, der humane Werte, das Menschenrecht auf Asyl, offensiv verteidigt. Und der nicht opportunistisch Positionen der #NoAfD übernimmt.

From anecdotal evidence: same holds for the German schools, I have contact with.

Scott McGrath@smcgrath.phd

Public records from early 2023 reveal US schools were largely unprepared for ChatGPT's rapid adoption by students. Responses varied widely, highlighting the lack of a unified strategy as the technology quickly became a significant challenge in education. #MLSky #AcademicSky #EduSky

Cunning idea by the looks of it!

Robert Haase@haesleinhuepf.bsky.social

Are you a #BioImageAnalysis pro? Challenge your skills by playing #BIAbubbles! 🎉 Find the right workflows and reproduce the result! 🔬🖥️🕹️ And if you have challenging BIA tasks, new level contributions are welcome! 😀 github.com/haesleinhuep...

Can’t wait to listen in!

Pierre-Simon Laplace@learnbayesstats.bsky.social

What if AI could know when it doesn’t know? 🎙️ @alex-andorra.bsky.social talks with @vincefort.bsky.social about Bayesian deep learning, why it matters for uncertainty, calibration & more 🎧 Tune in: learnbayesstats.com/episode/129-... #BayesianDeepLearning #MachineLearning #ReliableAI #AIResearch

psteinb.bsky.social reposted

Would you present your next NeurIPS paper in Europe instead of traveling to San Diego (US) if this was an option? Søren Hauberg (DTU) and I would love to hear the answer through this poll: (1/6)

docs.google.comNeurIPS participation in EuropeWe seek to understand if there is interest in being able to attend NeurIPS in Europe, i.e. without travelling to San Diego, US. In the following, assume that it is possible to present accepted papers ...

Great initiative to see if Neurips could be replicated in Europe! 🇪🇺

Serge Belongie@serge.belongie.com

Would you present your next NeurIPS paper in Europe instead of traveling to San Diego (US) if this was an option? Søren Hauberg (DTU) and I would love to hear the answer through this poll: (1/6)

It's been a blast, thanks to @sbi-devs.bsky.social ! This week's hackathon was phenomenal! 🙏 😍 The sbi hackathon welcomed about 25 people in Tübingen with contributions spanning the globe , e.g. 🇺🇸🇯🇵🇧🇪🇩🇪. Wanna see, what we did? Check out the PRs👇 github.com/sbi-dev/sbi/...

github.comPull requests · sbi-dev/sbisbi is a Python package for simulation-based inference, designed to meet the needs of both researchers and practitioners. Whether you need fine-grained control or an easy-to-use interface, sbi has ...

What a gem! Lovely article on our blind focus on accuracy in ML. Can't wait to get my hands on the references therein.

Working with #julia sometimes feels like playing minesweeper. Found a helpful library on the web 💣, punched in `add AwesomeLib` and boom 💥 I have to wait for 15min 🥱 to get all dependencies compiled and installed. #21century #oldhardware

I think, this is a great opportunity for anyone interested!

Mathieu Seyfrid@mathieuseyfrid.bsky.social

Thrilled about the launch of our first Helmholtz AI 🇩🇪x Mila 🇨🇦 call, for PhD students of Helmholtz AI Associates to visit groups at @mila-quebec.bsky.social and pollinate new connections with the @www.helmholtz.de Association: cross-field, cross-center. @helmholtzai.bsky.social

Jaw dropping results! Congrats @maxxxzdn.bsky.social for this wonderful treat. Happy to put this into play.

Max Zhdanov@maxxxzdn.bsky.social

🤹 Excited to share Erwin: A Tree-based Hierarchical Transformer for Large-scale Physical Systems joint work with @wellingmax.bsky.social and @jwvdm.bsky.social preprint: arxiv.org/abs/2502.17019 code: github.com/maxxxzdn/erwin

Welcome! 🤗

HZDR@hzdr.bsky.social

Hallo Bluesky! Wir sind das Helmholtz-Zentrum Dresden-Rossendorf (HZDR) und Teil der #Helmholtz -Gemeinschaft. Wir forschen auf den Gebieten #Energie, #Gesundheit und #Materie an den Herausforderungen unserer Zeit. Folgt diesem Kanal für spannende Einblicke und Neuigkeiten aus dem #HZDR.

Great to see flow matching more and more in the wild!

Paul Hagemann @yungbayesian.bsky.social

Our paper "PnP-Flow: Plug-and-Play Image Restoration with Flow Matching" has been accepted to ICLR 2025. Here a short explainer: We want to restore images (i.e., solve inverse problems) using pretrained velocity fields from flow matching. However, using change of variables is super costly.

psteinb.bsky.social reposted

Our paper "PnP-Flow: Plug-and-Play Image Restoration with Flow Matching" has been accepted to ICLR 2025. Here a short explainer: We want to restore images (i.e., solve inverse problems) using pretrained velocity fields from flow matching. However, using change of variables is super costly.

Great plot! 😍 A bit of clarification for the top entries: cs.CV = computer science->Computer Vision & Pattern Recognition cs.LG = CS->Machine Learning cs.CL = CS->Computation & Language quant-ph = Quantum Physics cs.RO = CS->Robotics cs.AI = CS->Artificial Intelligence

Thomas Dietterich@tdietterich.bsky.social

ArXiv continues to grow. Here is the year-on-year comparison for the categories with 1000+ submissions. Overall, 17% growth in submissions from 2023 to 2024 (208,493 -> 244,031)

This was a great episode that I loved very much. @khinsen.net I'd be great to know the repositories behind the MOOCs! I'd love to have a closer look. #reproducibility

Code for Thought / Peter Schmidt@code4thought.bsky.social

English edition [EN] #code4though: there is a gap in teaching post-grads how to make their work reproducible. In France, 3 researchers got together and created an online course to fill this gap. Listen to my conversation with them on your podcast app, YouTube podcast or codeforthought.buzzsprout.com

Totally on my reading list! 👇👇👇🤩👇👇👇

Mathurin Massias@mathurinmassias.bsky.social

Anne Gagneux, Ségolène Martin, @quentinbertrand.bsky.social Remi Emonet and I wrote a tutorial blog post on flow matching: dl.heeere.com/conditional-... with lots of illustrations and intuition! We got this idea after their cool work on improving Plug and Play with FM: arxiv.org/abs/2410.02423

psteinb.bsky.social reposted

The sbi package is growing into a community project 🌍 To reflect this and the many algorithms, neural nets, and diagnostics that have been added since its initial release, we have written a new software paper 📝 Check it out, and reach out if you want to get involved: arxiv.org/abs/2411.17337

arxiv.orgsbi reloaded: a toolkit for simulation-based inference workflowsScientists and engineers use simulators to model empirically observed phenomena. However, tuning the parameters of a simulator to ensure its outputs match observed data presents a significant challeng...