We are pleased to announce that the 2nd Workshop on Learning Under Weakly Structured Information will be held at the Tübingen AI Center.
After the first edition in 2024, organized by Hannah Blocher and Christoph Jansen from LMU Munich, the FMLS group at the University of Tübingen will host this theory-focused workshop for the current year.
About the Workshop
When: 7th and 8th of April 2025
Where: 7th: Room 08 (EG), Münzgasse 11, 72070 Tübingen
8th: Lecture Hall (EG), Maria-von-Linden-Straße 6, 72076 Tübingen
The workshop is meant as a small and interactive event. It will cover topics in the Theory of Machine Learning field, which usually fly below the radar of the fast-paced field. In particular, the topics include (but are not limited to):
- Uncertainty quantification
- Information theory in machine learning
- Imprecise probabilities in statistics and machine learning
- (Generalized) Bayesian methods
- Foundations of statistical and machine learning
- Philosophical aspects of machine learning
- Decision theoretic approaches to machine learning
The main organizers are Laura Iacovissi, Armando Cabrera Pacheco, and Rabanus Derr.
Program and registration
-> Registration deadline: March 28th, through this form.
Participation is free of charge. However, registration for the workshop is required to guarantee a smooth organization.
The workshop program includes a social event on the first night (dinner, dutch-treat basis), and plenty of time for engaging in scientific discussion during breaks.
-> Detailed Schedule:
|
MONDAY |
Title |
Speaker |
|---|---|---|
|
9:00 |
Hellolo |
|
|
9:30 - 9:40 |
Welcome |
|
|
9:45 - 10:25 |
What can post-hoc explanation methods tell us about our data? |
|
|
10:30 - 11:00 |
The Rhetorics of ML |
|
|
11:00 - 11:30 |
Break |
|
|
11:30 - 12:10 |
Towards Reciprocal Learning Theory: Generalizing From Self-Selected Samples |
|
|
12:10 - 13:30 |
Lunch |
|
|
13:30 - 14:10 |
Beyond $\mu$P: Scaling Insights from Infinite-width Theory for Non-standard Architectures and Learning Paradigms |
|
|
14:15 - 15:00 |
Size vs Capacity (or better Capacity vs Diversity?): On some subtle issues in highly nonparametric resampling-based tests that make use of VC-guarantees for regularization |
|
|
15:00 - 15:30 |
Break |
|
|
15:30 - 16:10 |
On Calibration in Multi-Distribution Learning |
|
|
16:15 - 17:00 |
Four Facets of Forecast Felicity - Revisited |
|
|
Dinner n Drinks (Dutch treat) |
|
TUESDAY |
Title |
Speaker |
|---|---|---|
|
9:00 |
Hellolo v2 |
|
|
9:30 - 10:10 |
Empirical Decision Problems |
|
|
10:15 - 11:00 |
Prediction, Potential Outcomes, and Performativity |
|
|
11:00 - 11:30 |
Break |
|
|
11:30 - 12:10 |
Data Processing Inequalities with Constrained Model Class |
|
|
12:10 - 13:30 |
Lunch |
|
|
13:30 - 14:10 |
TBA |
Nan Lu |
|
14:15 - 15:00 |
Correlation Uncertainty |
|
|
15:00 - 15:30 |
Break |
|
|
15:30 - 16:00 |
TBA |
Mohammad Amin Charusaie |
|
16:05 - 16:45 |
The Fairness-Quality Trade-off in Clustering |
Ana-Andrea Stoica |
|
FINISH! |
(Background image generated by ChatGPT)