ASCAI MEETING
Orsay - June 19-20, 2025
- The interface between statistical physics and learning.
- Computational barriers in statistical learning.
Organized at Institut de Mathématiques d'Orsay by the franco-german research group ASCAI.
The two days will be structured around two mini-courses and two advanced lectures. They will be preceded by a satellite workshop presenting some recent achievements of the ASCAI group.
Speakers / Schedule
Wednesday, June 18, Satellite meeting of ASCAI group.
Presentation of some recent achievements of the ASCAI research group. Room 3L15.
9h00 - 9h30 -- Room 1F12 : Welcoming coffee - viennoiseries & registration
9h30 - 10h30 -- Debarghya Ghoshdastidar (TUM) Statistical Guarantees for Gradient-based Optimisation: Transformers and Projection Pursuit.
Coffee break: room 1F12
11h00 - 11h35 -- Romain Périer (Orsay) Post hoc methods for heterogeneous tests.
11h35 - 12h10 -- Ulysse Gazin (Paris) Selection of informative conformal prediction regions.
Lunch
14h00 - 14h35 -- Maximilian Graf (Potsdam) Statistical Inference for Paired Spatial Poisson Processes with Missing Data.
14h35 - 15h10 -- Bertrand Even (Orsay) Low-degree Lower Bounds for models with Latent Variables; Application to Clustering, Sparse Clustering and Bi-clustering.
15h10 - 15h45 -- Victor Thuot (INRAE) Clustering Items through Bandit Feedback: Finding the Right Feature out of Many.
15h45 - 16h20 -- Sebastian Vogt (TUM) Non-Parametric Active Bandit Clustering.
Coffee break: room 1F12
17h00 - 18h00 -- Alexandra Carpentier (Potsdam) A simple algorithm for noisy, zeroth-order convex optimisation.
18h00 - … : discussion time
Thursday, June 19, room 3L15
9h00 - 9h30 -- Room 1F12: Welcoming coffee - viennoiseries & registration
09h30-11h00 -- Antoine Maillard. Interplay between statistical physics and learning (I)
Coffee break: room 1F12
11h30-12h30 -- Pierfrancesco Urbani. Generalization and overfitting in large overparametrized two-layer neural networks.
Lunch: room 1F12
14h00-15h30 -- Antoine Maillard. Interplay between statistical physics and learning (II)
Coffee break: room 1F12
16h00-17h30 -- Alex Wein. Computational barriers in statistical learning (I)
Friday, June 20, room 3L15
09h30-10h30 -- Elisabetta Cornacchia. Learning with neural networks: Complexity and Data Structure
Coffee break: room 1F12
11h00-12h30 -- Alex Wein. Computational barriers in statistical learning (II)
Lunch: room 1F12
Venue
Institut de Mathématique d'Orsay, université Paris Saclay (how to get there?) Room 3L15, Bâtiment 307, 307 Rue Michel Magat, 91400 Orsay.
Inside the building :
Room 1F12: enter by the main entrance of the building; take the small staircase in front of you, slightly on the right; go along the inner garden until you reach the room 1F12 on your left.
Room 3L15: go to room 1F12; take the staircase in front of this room, up to the 3rd floor; the room 3L15 is then on your left.
Organizers / Sponsors
Local organizers: Gilles Blanchard, Elisabeth Gassiat, and Christophe Giraud.
Funding: ASCAI DFG-ANR research group