New blog post: Skills for the Future A brief reflection on what skills may (hopefully) remain relevant in the new (hopefully bright) future of LLMs et al. #AcademicWriting #AcademicChatter bastian.rieck.me/blog/2025/sk...

Interesting application of ideas from topology to LLM training!

EurIPS student pricing is ending soon and I have a hunch that the other tickets might also sell out, so get yours now while supplies last!

It's epoch 50. Do you know whether your LLM is grokking? In our #NeurIPS2025 paper "Less is More: Local Intrinsic Dimensions of Contextual Language Models," we study how the local intrinsic dimension (LID) of embeddings changes during training and fine-tuning. #research 🧵1/n

Want to generate point clouds? Our (w/ Ernst Röell) #NeurIPS2025 paper shows a topology-driven method that is fast, simple, and high-quality. 🔥 1/n

Oh, you have criticisms of my paper? Surely you noticed the part of the paper where we already rebutted your critique, by writing “Possible criticisms: the methodology of this paper is bad. We are aware of this.”

If you teach in higher education: You may be grappling with the existence of generative AI tools. I’ve landed on “they’re not going away, so I educate students and have them co-create their own class policy for use.” They do a wonderful job, and I once wrote up the process and results. #AcademicSky

Currently, on Hacker News.

Will AI Replace Human Thinking? The Case for Writing and Coding Manually.

(if you're really curious, we have a full github repo for the paper here: github.com/dkobak/low-d...)

one of the more fun topics stemmed from a discussion with @hippopedoid.bsky.social over what the oldest 2D PCA visualization we could find was. After some scouring, we settled on a 1960 paper from researchers at the Université de Montréal about turtle carapaces, which we recreated.

Last year I met a bunch of great researchers who work with high-dimensional data at a Dagstuhl seminar. This week we put out a preprint about the history and philosophy of low-dimensional embedding methods, their applications, their challenges, and their possible future arxiv.org/abs/2508.15929

Hope you all had a good summer. I'm very happy to announce the speaker line-up for the falls Chalmers AI4Science seminars! Hope to catch you all there!

New blog post: Consciousness Is Overrated Anyway (for AI) In which I, somewhat coherently, try to contribute to the whole "Is AI Conscious" debate. 🔗 bastian.rieck.me/blog/2025/co... #AI #MachineLearning

Should we still teach machine learning classes? Let me justify my existence.

Wonder when I'll reach the stage of my career where I just sign with an initial... People who do that already, how cool do I have to be to join the club? 🥺 #AcademicSky #MathSky

Come find me (and this lovely bin of recyclable pens) later today at 181! @cogcompneuro.bsky.social

Save the Date! ATMCS 12 is coming up fast, January 26-30, 2026, in Leipzig. More info at the link below, call for contributions coming soon! www.mis.mpg.de/events/serie...

📢 Save the date! Join us for the next @ellis.eu x UniReps Speaker Series! 📅 27th August – 16:00 CEST 📍https://ethz.zoom.us/j/66426188160 🎙️ Speakers: Keynote by @lelandmcinnes.bsky.social & Flash Talk by Yu (Demi) Qin 🔔 Stay updated by joining our Google group: groups.google.com/u/2/g/ellis-...

🎥 Shapes, Spaces, Simplices, and Structure: Geometry, Topology & Machine Learning "What if the answer to some problems in graph learning is not more, but better structure?" This is the central premise of my talk @logml.bsky.social and @unireps.bsky.social. 🧵1/5

"SAFE TY FIRST" Kerning. Once you know about it, your enjoyment of public signage will be ruined forever. en.wikipedia.org/wiki/Kerning

Happy to announce I just defended my dissertation at Harvard and I’m now a PhD.

🇨🇭 Happy #SwissNationalDay! It's already a year since I started my new position @unifr.bsky.social and I'v had the good fortune to be living here since 2018. Thanks for everything, Switzerland! 🙏 Here's to the future! 🚀

While you are waiting for your #NeurIPS reviewers, you might enjoy this tongue-in-cheek blog post I wrote: The Missing Screwtape Letters from Academia bastian.rieck.me/blog/2023/sc... (With apologies to C.S. Lewis)

We grow better and stronger: a few inspiring visits between Mila and Helmholtz Munich: 1. read below on what a few week stays at Mila brought to Irene and Gab, PhD students from Benjamin Schubert and Maria Colome Tatché 's groups @HelmholtzMunich […] [Original post on wisskomm.social]

Looking very much forward to this event! Thanks for having me!

🔥 Mark your calendars for the next session of the @ellis.eu x UniReps Speaker Series! 🗓️ When: 31th July – 16:00 CEST 📍 Where: ethz.zoom.us/j/66426188160 🎙️ Speakers: Keynote by @pseudomanifold.topology.rocks & Flash Talk by @florentinguth.bsky.social

Topology is cool, check it out!

#MachineLearning loves data. #Topology loves shape. Let's bring them together! 👉The Euler Characteristic Transform (ECT) bridges both worlds, as I outline in a recent article in the Notices of the @amermathsoc.bsky.social Engage! 🧵1/n

Join us on a (mathematical) journey to a shire - oops, HIGHER - standard and principled evaluation schema for our benchmark datasets. This is the reward of the RINGS framework. 📒 Blog: aidos.group/blog/rings/ 📃 Paper: doi.org/10.48550/arX... 👩‍💻 Code: github.com/aidos-lab/ri...

✨Excited to be at ICML 2025 to share our new evaluation framework for graph-learning datasets, work with @dataspider.bsky.social, @jeremy-wayland.bsky.social, @pseudomanifold.topology.rocks Curious about who shall pass, and who shall not?🤔 Enjoy nods to LoTR?🧝‍♀️ Check out our blog and paper below!

Next appointment: 31st July 2025 – 16:00 CEST on Zoom with 🔵Keynote: @pseudomanifold.topology.rocks (University of Fribourg) 🔴 @florentinguth.bsky.social (NYU & Flatiron)

A great slide from a great talk by @pseudomanifold.topology.rocks at @logml.bsky.social

📢 Call for Papers: NeurReps 2025 ‼️‼️‼️ 🧠 Submit your research on symmetry, geometry, and topology in artificial and biological neural networks. Two tracks: Proceedings (9 pages) and Extended Abstract (4 pages). Deadline: Aug 22, 2025. www.neurreps.org/call-for-pap...

🚨One of the coolest workshops in AI is back! NeurReps 2025 is calling for papers on symmetry, geometry & topology in neural networks 🧠📐 If your work bridges theory & structure — don’t miss this. 📅 Deadline: Aug 22

Very happy that Jiří Minarčík will join our research group at @cmu.edu, the Geometry Collective, as a Fulbright visiting scholar! 🥳 Jiří is a world expert in space curves, and one of the core contributors to Penrose (penrose.cs.cmu.edu). Check out his beautiful work here: minarcik.com

I am very happy to report that two of these students will join the AIDOS Lab @unifr.bsky.social to work on #geometry, #topology, and #MachineLearning 😊

Check out this post from @ffabffrasca.bsky.social and the GLOW reading group on the future of graph learning! I’ve also contributed and my main take is - its actually working and its an exciting moment to work on applications! open.substack.com/pub/glowread...

You may read this chapter in the usual place: geometricdeeplearning.com/book/ 📖 Any and all feedback / thoughts / questions on the content... words of encouragement for finishing this book (pretty please! 😇) warmly welcomed! @mmbronstein.bsky.social @joanbruna.bsky.social @taco-cohen.bsky.social

The first draft 'G' chapter of the geometric deep learning book is live! 🚀 Alice enters the magical, branchy world of Graphs and GNNs 🕸️ (LLMs are there too!) I've spent 7+ years studying, researching & talking about graphs -- This text is my best attempt at conveying everything i've learnt 💎

Today KellyMaggs @k311ym4gg5.bsky.social @math-mpicbg.bsky.social @mpicbg.bsky.social @tudresden.bsky.social will give talk on mathML research! @krishnaswamylab.bsky.social

new collaborator

Kelly Maggs @k311ym4gg5.bsky.social, a postdoc in the Algebraic Systems Biology group, thinking about ways of coming from mathematics and approaching interdisciplinary work with biologists

Friends at #AcademicSky, I wrote a new blog post on "clopen data" (sic!), collecting my thoughts on how #MachineLearning should evolve to value "data production," while also calling out the bad practice of "not-quite-open data." Hence: clopen! bastian.rieck.me/blog/2025/cl...

What are the geniune benefits of the PI model if the PI is inventivized to just do little else other than thought leadership given the literal endless supply of strong students? What is it meant to achieve beyond [an undeserved] Matthew effect for scoring in the glorified nepotism lottery?

Very happy that the grant proposal I wrote got accepted by the Swiss Science Foundation. Together with Deborah Oliveros and Bernd Gärtner we will study aspects of Tverbergs theorem. The project also funds a postdoc in Mexico and a PhD student at ETH. Looking forward to mentoring my first PhD student

After a break in April we welcome all of you to the next session of GLOW🌟 next week! Join and interact with our speakers Christian Koke (scale in GNNs) and Yonatan Sverdlov (sparse geometric MPNNs). 🗓️ May 28th, 5pm CEST on Zoom. 🌐 Details & sign-up: sites.google.com/view/graph-learning-on-weds.

TDL actually works at scale! And we believe 𝐇𝐎𝐏𝐒𝐄 lays the foundation for broad applications of TDL ✨ 📭 Reach out for collaborations Special thanks to @levtelyatnikov.bsky.social and @gbg1441 and the team @ninamiolane.bsky.social and @Coerulatus :) (6/6)

Empirically we find that 𝐇𝐎𝐏𝐒𝐄 👀: ⚡Outperforms (speed and performance) GNNs in larger datasets (MANTRA) 🥇Achieves SOTA on Topological Tasks -- predicting Betti numbers 1 & 2 on MANTRA 📈 Is: faster, more accurate, scalable (5/6)

The model relies on positional and structural encodings (PSE) of the combinatorial domain. By decomposing a combinatorial representation into neighborhoods which take the form of a Hasse graph, the PSE of each can be used to discard the topology and learn from the PSE! (4/6)