Stefan Neumann | Lightnews

Stefan Neumann @neumannstefan.com · Sep 12

I really like this setting of "given a suboptimal solution, improve it". It feels natural since data scientists often start with good solutions & then refine them.

Led by the great @sijingtu.bsky.social and joint with Aleksa Stankovic and @aris-gionis.bsky.social.
link.springer.com/article/10.1...

Optirefine: densest subgraphs and maximum cuts with k refinements - Data Mining and Knowledge Discovery

Data-analysis tasks often involve an iterative process, which requires refining previous solutions. For instance, when analyzing social networks, we may obtain initial communities based on noisy metad...

link.springer.com

Stefan Neumann @neumannstefan.com · Sep 12

Further, we give approximation algorithms for this problem, based on semidefinite programming and the sum-of-squares hierarchy.

We also prove lower bounds showing that our approximation ratios are close to what one can hope for.

Besides MaxCut we provide similar results for Densest Subgraph.

Stefan Neumann @neumannstefan.com · Sep 12

Interestingly, we show that there are close relationships of this problem to cardinality-constrained versions of the problem.

We show that even black-box reductions are possible and that they work surprisingly well in experiments.

Stefan Neumann @neumannstefan.com · Sep 12

Concretely, suppose you are given a cut in a graph. This cut partitions the graph's nodes into two sides.

Now you are allowed to move k nodes from one side of the partition to the other.

Which k nodes should you use to increase the cut as much as possible?

Stefan Neumann @neumannstefan.com · Sep 12

Suppose you are given a non-optimal solution to an algorithmic problem and you want to make a small number of changes to improve its objective function.

We give nearly-optimal approximation algorithms for this problem for Max-Cut and for Densest Subgraph.

link.springer.com/article/10.1...

Optirefine: densest subgraphs and maximum cuts with k refinements - Data Mining and Knowledge Discovery

Data-analysis tasks often involve an iterative process, which requires refining previous solutions. For instance, when analyzing social networks, we may obtain initial communities based on noisy metad...

link.springer.com

Stefan Neumann @neumannstefan.com · Sep 9

TU Wien is hiring for a Tenure-Track Assistant Professor of Information and Communication Technologies in Automation.

This is your chance to join a wonderful Top-13 computer science department in Europe (based on CSRankings).

Contact me if you want to chat.

informatics.tuwien.ac.at/news/2967

Assistant Professor of Information and Communication Technologies in Automation

We invite applications for a full-time assistant professorship with tenure track in Information and Communication Technologies in Automation until Nov 13, 2025.

informatics.tuwien.ac.at

TCS+ @tcsplus.bsky.social · Sep 8

With September (already!) underway, the next season of TCS+ will start soon. As we are planning for it, a reminder that you can—and are encouraged to!—submit suggestions for talks and speakers on our website: sites.google.com/view/tcsplus...

TCS+ - Suggest a talk

Suggest a talk

sites.google.com

Stefan Neumann @neumannstefan.com · Sep 4

You just need a European host university to apply. Of course then you have to move to Europe if you get the grant and want to accept it. But I guess that many people are considering to do this right now.

Stefan Neumann @neumannstefan.com · Sep 4

I thought about it and here is my conjecture: LLMs have made grant writing a lot more efficient, so there are just a lot more people submitting.

That would also explain why there was a significant increase already two years ago and not just this year (which the US situation could not explain).

Stefan Neumann @neumannstefan.com · Sep 2

Wow. I wonder what triggered the rise in submissions in the past two years?

“The ERC received 3 329 proposals — 31% more than last year and 82% more than the year before that.”

Stefan Neumann @neumannstefan.com · Sep 2

This triggered a lot of follow-up research in the theory community (see link below), under the name of "algorithms with predictions". However, my impression is that few of the algorithms are practical, due to unrealistic assumptions on oracles (&other reasons).

algorithms-with-predictions.github.io

ALPS

algorithms-with-predictions.github.io

Paper Skygest Team @paper-feed.bsky.social · Aug 19

**Please repost** If you're enjoying Paper Skygest -- our personalized feed of academic content on Bluesky -- we'd appreciate you reposting this! We’ve found that the most effective way for us to reach new users and communities is through users sharing it with their network

Stefan Neumann @neumannstefan.com · Aug 15

It depends a lot on what „understands“ means here. One could make a similar point about optimization problems: Simple heuristics (like greedy algorithms) work very well in practice, but the underlying problems are often hard to approximate and we typically have no formal explanation for this gap.

Stefan Neumann @neumannstefan.com · Aug 12

We all want to make the review process as fair and thorough as possible. But I don't think that adding more pressure on reviewers and generating even more text is the way.

(Also, my personal impression is that the review quality at NeurIPS/ICML is a lot better than its reputation.)

Stefan Neumann @neumannstefan.com · Aug 12

By now it's just too much information (reviews, rebuttal back-and-forth, final reviewer justification, final author remark).

On average, for each paper I'm reviewing there is a discussion of ~10 pages.

It's almost impossible to keep track of everything.

Rémi Flamary @rflamary.bsky.social · Aug 11

Why #NeurIPS PC? Why? We already have days of reading to do including reviews, responses, responses to responses and confidential comments to AC. Do you really think a "Final Remark" that tells me to accept the paper will help? Let me just do my AC job during august while everyone is on vacation!

Rémi Flamary @rflamary.bsky.social · Aug 11

Why #NeurIPS PC? Why? We already have days of reading to do including reviews, responses, responses to responses and confidential comments to AC. Do you really think a "Final Remark" that tells me to accept the paper will help? Let me just do my AC job during august while everyone is on vacation!

Stefan Neumann @neumannstefan.com · Aug 5

I am also deeply thankful to Wolfgang Dvořák and our great team of senior lecturers and student assistants, who ensure the high quality of exercise sessions and that everything goes smoothly. This is a great achievement in a course with more than 700 students.

Stefan Neumann @neumannstefan.com · Aug 5

The course “Fundamentals of Digital Systems” that I am co-teaching has been nominated for the Best Lecture Award at TU Wien (among the best 9 courses of the entire university!).

This is a great honor and I am so grateful to our students for nominating our course.

www.tuwien.at/tu-wien/aktu...

Das sind die Best Teaching Award-Finalist_innen 2025

Die Jury hat entschieden, für wen die Eulenjagd weitergeht.

www.tuwien.at

VCLA | TU Wien @vclatuwien.bsky.social · Aug 5

❗️ JOB OFFER! University Assistant (post-doc)
Institute of Logic and Computation at @tuwien.at | @informatics.tuwien.ac.at

Research Unit Formal Methods in Systems Engineering
full-time, starting in September and limited to one year
🗓️ DL August 14
ℹ️ jobs.tuwien.ac.at/Job/255265
#postdocposition

University Assistant Post-Doc (all genders)

jobs.tuwien.ac.at

AITHYRA Research Institute of the Austrian Academy of Sciences @aithyra.bsky.social · Aug 3

Postdoctoral Position in Geometric Deep Learning & AI for Science (f/m/d)

Michael Bronstein, AITHYRA Director and Hon Prof TU Wien in collaboration with Ismail Ilkan Ceylan, expert in graph ML, invites to APPLY NOW: aithyra.onlyfy.jobs/job/ltt3a9j6
Deadline 31.8.2025

#AITHYRA #TUWien #PostDoc

Michael Bronstein in Vienna, St. Stephen's Cathedral in the back (c) Natascha Unkart

Max-Planck-Gesellschaft @maxplanck.de · Aug 1

Director at Max Planck - a unique position! The Open Call for Expressions of Interest in Max Planck Directorships is open now and can be submitted by the 31st of October 2025. ➡️ mpg.de/directors - Please share the Open Call among potential candidates.

Open Call for Expressions of Interest in Max Planck Directorships:
Expressions of interest can be submitted until 31 October 2025.

Stefan Neumann @neumannstefan.com · Jul 31

I don’t know about downranking, but we once tried to study how newspaper articles (=links) are discussed on social media. We looked at how the same article is discussed across different communities on the same platform. Unfortunately, the data was too sparse, so we abandoned the project.

Johan Ugander @jugander.bsky.social · Jul 30

Does anyone know studies of how FB/X/threads/etc downrank (or don't) external links? According to Adam Mosseri in 11/2024, threads then didn't explicitly downrank links, but notes threads prioritizes like/comments, not clicks, and says people simply don't like/comment much on posts with links.

Mosseri: We don't downrank links, but we don't place much value on p(click) and people don't like and comment on links much.

Danupon @nanongkai.bsky.social · Jul 30

📢 Full Professor (W3) in Algorithms & Complexity @ Saarland University 🇩🇪

Join a strong TCS environment with MPIs, CISPA & more!

🗓 Deadline: Sep 18, 2025
📄 Details: www.uni-saarland.de/fileadmin/up...
📝 Apply: www.uni-saarland.de/verwaltung/b...
📬 Contact: mblaeser@cs.uni-saarland.de

This link will take you to a page that’s not on LinkedIn

lnkd.in

Stefan Neumann @neumannstefan.com · Jul 28

Letzte Woche habe ich in der Sendung „Spontan gefragt“ auf KURIER TV über meine Forschung gesprochen. Es ging darum, wie Algorithmen unsere Wahrnehmung in sozialen Medien beeinflussen – von der Auswahl von Inhalten bis zur Meinungsbildung.

@kurier.at @wwtf.at

TU Wien @tuwien.at · Jul 25

Was entscheidet, was wir auf Social Media sehen?
Heute bei „Spontan gefragt“ auf KURIER TV: TUW-Forscher @stefanresearch.bsky.social erklärt, wie Algorithmen unsere digitale Realität formen.
🕘 21:15 auf KURIER TV
📺 Mediathek: kurier.tv/spontan-gefr...

TU Wien @tuwien.at · Jul 25

Was entscheidet, was wir auf Social Media sehen?
Heute bei „Spontan gefragt“ auf KURIER TV: TUW-Forscher @stefanresearch.bsky.social erklärt, wie Algorithmen unsere digitale Realität formen.
🕘 21:15 auf KURIER TV
📺 Mediathek: kurier.tv/spontan-gefr...

Manoel Horta Ribeiro @manoelhortaribeiro.bsky.social · Jul 22

🚨 Study alert!

We've developed a prototype that enables you to create personalized social media feeds using natural language rules and import them into Bluesky.

Join our study to try out the tool, give us your feedback, and earn a $50 gift card!

Sign up: forms.gle/MkSGKzxDfBEc...

Please share :)

Ingmar Weber @ingmarweber.de · Jul 18

🚨Job Alert
W2 (TT W3) Professorship in Computer Science "AI for People & Society"
@saarland-informatics-campus.de/@uni-saarland.de is looking to appoint an outstanding individual in the field of AI for people and society who has made significant contributions in one or more of the following areas:

Stefan Neumann @neumannstefan.com · Jul 17

Very nice. I might mention this to my first semester students when I tell them about floating point numbers and finite precision.

Light upyour news

Sign in to Lightnews

Sign up to start reading

Connect Bluesky

Connect with Bluesky

Light up
your news