Humboldt-Universität zu Berlin - Mathematisch-Naturwissenschaftliche Fakultät - Institut für Mathematik

Berlin-Bielefeld-Paris Workshop on Early Stopping

April 13-14, 2023 in Berlin

 

 

A2_Mathematik_2023_V2.jpg

 

Organisers

Gilles Blanchard (Université Paris-Saclay)
Alain Celisse (Université Paris 1 Panthéon-Sorbonne)
Markus Reiß (Humboldt-Universität zu Berlin)
Martin Wahl (Universität Bielefeld)

 

Venue

Erwin-Schrödinger-Zentrum
Room 1'303
Rudower Chaussee 26, 12489 Berlin

 

Hotel

Airporthotel Berlin-Adlershof
Rudower Chaussee 14
12489 Berlin

 

Programme

 

Abstracts

 

Thursday, April 13

09:00-09:50

Patrick Rebeschini

Implicit regularization in statistical learning: An overview

09:50-10:20

Coffee Break

10:20-11:10 

Bernhard Stankewitz

Early stopping for $L^{2}$-boosting in sparse high-dimensional linear models

11:10-12:00

Alain Celisse

Early stopping and regularization by projection for spectral filter estimators

12:00-14:00 Lunch Break
14:00-14:50

Garvesh Raskutti

Early stopping and regularization: From kernel methods to variable importance in neural networks

14:50-15:30

Coffee Break

15:30-16:20

Sylvain Arlot

Cross-validation for estimator selection

16:20-17:00

Perspectives

18:30-

Workshop Dinner 

 

Friday, April 14

09:00-09:50

Laura Hucker and Markus Reiß

Early stopping for conjugate gradients in statistical inverse problems

09:50-10:40

Tim Jahn

Early stopping of untrained convolutional networks

10:40-11:10 Coffee Break
11:10-12:00

Nicole Mücke

Learning linear operators: Infinite-dimensional regression as a well-behaved non-compact inverse problem

12:00-13:00

Lunch 

 

 

Dinner

Transit
Sonntagstraße 28,
10245 Berlin

 

Participants

Interested researchers are welcome to participate. Due to restricted capacities registration by email to martin.wahl@math.hu-berlin.de is mandatory.

 

 

Contact

martin.wahl (at) math.hu-berlin.de

logo_hu.png   640px-Universität_Bielefeld_Logo.svg.png   logo_coul_fr.png   Logotype UPSaclay_CMJN.jpg