Ulrike Luxburg
@UlrikeLuxburg
Followers
3K
Following
46
Statuses
59
Professor of machine learning theory, University of Tübingen
Tübingen, Germany
Joined October 2020
Our exhibition about AI in Tübingen receives the Communicator Award! Thanks to Wiebke Ratzeburg and Guido Szymanska @StadtmuseumTue and to all the students who put so much effort in conceptualizing and creating the exhibition!
Der #Communicatorpreis 2024 geht an: Team "Cyber and the City"🎉@ulrikeluxburg, Tim Schaffarczik & Thomas Thiemeyer @uni_tue bringen m. einer Ausstellung das Thema #KI Menschen näher. Das Konzept kann modellhaft für kontrovers diskutierte Wissenschaftsthemen sein. Glückwunsch!
4
6
38
Women in machine learning in Tübingen, come join us on march 8!
🎉 Get Ready for the first TWiML Workshop! 🎉 📅 Save the Date: March 8th, 2024 📅 We are glad to invite you to the first TWiML - Tübingen Women in Machine Learning- event! 🌟 Everyone is Welcome! 🌟 @uni_tue @ml4science @MPI_IS
1
2
16
My postdoc @s_bordt started a blog - have fun reading his first post on LLMs and interpretability!
I've started a blog. The idea is to cover current topics in machine learning research. The first post asks: How will LLMs change interpretable machine learning?
0
2
1
Der #AIAct ist nicht perfekt, aber eine Zukunft ohne KI Regulierung wäre ein Disaster! Der AI Act muss JETZT verabschiedet werden! Er wird nie perfekt sein, dazu ist die Entwicklung in KI zu schnell. Bitte dafür stimmen! @Wissing @MarcoBuschmann @HabeckPress @OlschowskiPetra
2
0
7
@JessicaHullman Thanks Jessica, for your nice summary of my - indeed somewhat pessimistic - view on XAI.
1
1
5
RT @JessicaHullman: Blog post summarizing @UlrikeLuxburg's not-so-optimistic take on XAI at NeurIPS: we can make guarantees about conventio…
0
4
0
University of Tübingen is hiring ML professors on all levels, deadline is Jan 15. And we would love to increase the diversity of our team! Do apply!
Apply by January 15 to join the Tübingen AI faculty as a tenure-track or full professor! Want to learn more? Check out or connect with me at #NeurIPS2023 ☕
1
2
9
I am looking for a PhD student in explainable machine learning. You need strong mathematical skills (the goal is to formulate theorems and prove them) and an open mind when it comes to talking to applied ML, social science, ethics or law.
3
29
67
📷 Benign overfitting even works in small, fixed dimension! Not with ReLu activation function or smooth kernels, but with extra-spiky variants. #neurips2023 paper with @moritz_haa, @DHolzmueller, Ingo Steinwart:
2
13
32
Looking for independent research group leaders at our new institue for AI and Law in Tübingen: If you are a postdoc in law or AI, ready to establish your own research group, do apply! Deadline: Oct 15.
0
14
23
Our AI exhibition at @StadtmuseumTue got extended until January. If you are in Tübingen come visit, or check out the trailer in the blog:
0
4
9
Our AI exhibition in @StadtmuseumTue got extended until January! Also check out our blog about the process of designing it.
0
2
1
"Pitfalls of Climate Network Construction: A Statistical Perspective". With @moritz_haa and @bedartha, we show that many observations in climate networks might be spurious artifacts. And I am sure the same would hold for brain networks.
1
16
23
Three of our PhD students @solveigKlepper @janlause, Matthias Karlbauer, explaining to and discussing with journalists and the general public about "AI and sustainability", over course of the next months! (in german)
0
10
16
ChatGPT passed our algorithms exam for BSc computer science students (without prompt engineering, in a blind grading setup). GPT-4 is even better. Neat idea and implementation by @s_bordt
The default version of ChatGPT would have narrowly passed the exam with 20,5 out of 40 exam points. GPT-4 improves to 24 out of 40 exam points -- on par with the average student.
0
8
18
You want to understand which explanations SHAP produces for arbitrary functions? Or how to construct increasingly complex SHAP explanations for increasingly complex function classes? Read the answer in our #aistats23 paper with @s_bordt!
Are there connections between post-hoc explanation methods and interpretable models? In our #aistats23 paper, we show there is a strong correspondence between Shapley Values and Generalized Additive Models (GAMs). With @UlrikeLuxburg 1/N🧵
0
3
7
Our exhibition #cyberandthecity @stadtmuseumTue opened with 500 visitors! Come have a look if you are in Tue, it runs till october.
0
7
18
Our Exhibition about AI in @StadtmuseumTue opens this friday! Created jointly by 30 master students of anthropology and machine learning @uni_tue, over the course of more than a year.
1
22
39
I am looking for a postdoc, to work on the theory of explainable machine learning. More details:
3
54
169
My "Kinderuni" lecture goes online: Why is AI not always fair? Waiting for lots of kids to watch and ask live questions afterwards (in german). Live streaming this thursday at 17:00 (and then, forever on youtube).
0
10
20