Lightnews — Scholar-powered news

Bastian Grossenbacher-Rieck @pseudomanifold.topology.rocks · 8d

You publish papers that can be criticized. I submit to arXiv under an anonymous handle. We are not the same.

(/s)

Reposted by Bastian Grossenbacher-Rieck

Daniel Litt @littmath.bsky.social · 9d

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.”

5 14 90

Reposted by Bastian Grossenbacher-Rieck

Chad Topaz Queer DEI Race Traitor @chadtopaz.bsky.social · 27d

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

Empowering Students in the Age of AI

Co-Creating Responsible Use Policies

chadtopaz.medium.com

2 3 10

Reposted by Bastian Grossenbacher-Rieck

Simon Späti 🏔️ @ssp.sh · Aug 28

Currently, on Hacker News.

Will AI Replace Human Thinking? The Case for Writing and Coding Manually on the Front Page of Hacker News.

1 2 6

Reposted by Bastian Grossenbacher-Rieck

Simon Späti 🏔️ @ssp.sh · Aug 28

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

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

Learning to Think Again, and the Cost of AI Dependency. There are so many (hype/boring) posts about AI coming out every day. It’s OK to use it, and everyone does it, but still learn your craft, and tr...

www.ssp.sh

1 4

Reposted by Bastian Grossenbacher-Rieck

Alex Diaz-Papkovich @alexandr.bsky.social · Aug 27

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

GitHub - dkobak/low-dim-embeddings-review

Contribute to dkobak/low-dim-embeddings-review development by creating an account on GitHub.

github.com

1 5

Reposted by Bastian Grossenbacher-Rieck

Alex Diaz-Papkovich @alexandr.bsky.social · Aug 27

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.

Recreated PCA figure from: P. Jolicoeur and J. E. Mosimann. Size and shape variation in the painted turtle. A principal component analysis.
Growth, 24:339–354, 1960.

The figures show PC1 vs PC2 and PC2 vs PC3, with colours and symbols reflecting the sex of the turtle.

1 2 7

Reposted by Bastian Grossenbacher-Rieck

Alex Diaz-Papkovich @alexandr.bsky.social · Aug 27

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

The participants of Dagstuhl Seminar 24122 standing on steps outside (from https://www.dagstuhl.de/24122)

Multiple types of embeddings (UMAP, t-SNE, Laplacian Eigenmaps, PHATE, PCA, MDS) of Wikipedia text data labelled by a text summaries generated by an LLM. Methods like UMAP and t-SNE show cluster structure that reflect shared subject matter in text, whiel other methods show more continuous structure.

Multiple embedding methods (PCA, Laplacian Eigenmaps, t-SNE, MDS, PHATE, UMAP) of primate brain organoids at different time periods. Different methods highlight different aspects of development, such as clusters of similar cell types or time courses of cell development.

Multiple embedding methods (PCA, Laplacian Eigenmaps, t-SNE, MDS, PHATE, UMAP) of 1000 Genomes Project genotypes. Different methods reflect different aspects of demographic history of populations.

1 7 14

Reposted by Bastian Grossenbacher-Rieck

Simon Olsson @smnlssn.bsky.social · Aug 25

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!

2 8

Bastian Grossenbacher-Rieck @pseudomanifold.topology.rocks · Aug 25

...but there's a fundamental issue in terms of reporting and reproducibility culture. We need to make the community aware of these issues!

Let me know what you think :-)

3/3

1

Bastian Grossenbacher-Rieck @pseudomanifold.topology.rocks · Aug 25

A fair point by reviewers is to look for larger datasets (where LR is not going to scale) & add more baselines (always important when doing bake-offs).

Nevertheless: This should make us _pause_ and reconsider what we are doing.

We started a discussion (arxiv.org/abs/2502.023....

2/n

No Metric to Rule Them All: Toward Principled Evaluations of Graph-Learning Datasets

Benchmark datasets have proved pivotal to the success of graph learning, and good benchmark datasets are crucial to guide the development of the field. Recent research has highlighted problems with gr...

arxiv.org

1 1

Bastian Grossenbacher-Rieck @pseudomanifold.topology.rocks · Aug 25

I did! That's absolutely *bonkers*. Couple of thoughts already:

- Would love to see more performance comparisons (Platonov et al. present some great heterophilous datasets).

- Might be another nail in the coffin of Cora et al.

- TMLR comments are crazy openreview.net/forum?id=U0W...

1/n

Graph as a Feature: Improving Node Classification with Non-Neural...

Graph Neural Networks (GNNs) and their message passing framework that leverages both structural and feature information, have become a standard method for solving graph-based machine learning...

openreview.net

1 1

Bastian Grossenbacher-Rieck @pseudomanifold.topology.rocks · Aug 24

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

To briefly summarize my argument (cf. Orwell): When you are being stomped on by a boot, it does not matter whether the wearer is a conscious entity and capable of reflection. It just hurts.

1 1 4

Reposted by Bastian Grossenbacher-Rieck

Ben Recht @beenwrekt.bsky.social · Aug 22

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

Just When I Thought I Was Out

They pull me back in! Time to teach machine learning again.

www.argmin.net

5 4 37

Bastian Grossenbacher-Rieck @pseudomanifold.topology.rocks · Aug 16

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

2 5

Reposted by Bastian Grossenbacher-Rieck

Erica Busch @elbusch.bsky.social · Aug 15

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

1 8

Reposted by Bastian Grossenbacher-Rieck

Liz Munch @elizabethmunch.bsky.social · Aug 14

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...

Series

www.mis.mpg.de

1 3

Reposted by Bastian Grossenbacher-Rieck

UniReps @unireps.bsky.social · Aug 14

📢 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-...

6 10

Bastian Grossenbacher-Rieck @pseudomanifold.topology.rocks · Aug 8

Very interesting, talking a look… 👀

1

Bastian Grossenbacher-Rieck @pseudomanifold.topology.rocks · Aug 7

Feedback is always welcome. And thanks to everyone who helped me shape these thoughts into a coherent (w)hole! 🙏

🧵5/5

#MachineLearning #Geometry #Topology

1 1

Bastian Grossenbacher-Rieck @pseudomanifold.topology.rocks · Aug 7

If you are excited about representation learning, point clouds, higher-order complexes, or just curious about geometry/topology, this talk is for you!

👉 youtube.com/watch?v=frjm...

🧵4/5

Shapes, Spaces, Simplices, and Structure: Geometry, Topology, and Machine Learning

YouTube video by Bastian Grossenbacher-Rieck

youtube.com

1 1 4

Bastian Grossenbacher-Rieck @pseudomanifold.topology.rocks · Aug 7

Geometry and topology help us see structure and shape, providing a principled way to imbue models with such concepts…

…and it's not a niche perspective, thanks to a lot of great work by the community!

🧵3/5

The duality between geometry and topology

1 1

Bastian Grossenbacher-Rieck @pseudomanifold.topology.rocks · Aug 7

We all know the dogma:

"Given enough data, the model will learn everything."

But as LLMs hit diminishing returns (bough with even more GPUs!), this belief is being challenged.

I'm not the only one who posits that inductive biases remain important.

🧵2/5

1 2

Bastian Grossenbacher-Rieck @pseudomanifold.topology.rocks · Aug 7

🎥 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

1 1 9

Bastian Grossenbacher-Rieck @pseudomanifold.topology.rocks · Aug 6

"SAFE TY FIRST"

Kerning. Once you know about it, your enjoyment of public signage will be ruined forever.

en.wikipedia.org/wiki/Kerning

2