Context and objectives

The workshop on "Challenging problems in Statistical Learning" will be held in Lille on April 5-6, 2012. This workshop aims to summarize the new and future problems in statistical learning and to give a good idea of what already exists for dealing with these problems. This event is co-organized by the University Paris 1 Panthéon-Sorbonne, INRIA Lille - Nord-Europe, l'équipe Inria Modal and the Laboratoire Paul Painlevé.

Registration is free of charge but mandatory by March 15, 2012. You can register using the Registration page. Poster submission must be done by February 29, 2012 using the Poster registration page.

Thursday April 5, 2012

  • Session 1: Statistical learning and vizualization.
    • 09:30. Luca Scrucca (Università degli Studi di Perugia)
      • Dimension reduction based on finite mixture modeling of inverse regression
    • 10:30. Coffee break
    • 11:00. Amaury Lendasse (Aalto University, School of Science and Technology)
      • Information Visualization : An Introduction to the Field and Applications for Statistics Titre à préciser
    • 12:00. Petra Isenberg (INRIA Saclay, team Aviz)
      • Information Visualization : An Introduction to the Field and Applications for Statistics
  • Session 2: Statistical learning in high dimension.
    • 14:30. Arthur Gretton (University College London)
      • Hypothesis Testing and Bayesian Inference: New Applications of Kernel Methods
    • 15:30. Coffee break
    • 16:00. Sophie Dabo-Niang (University Charles De Gaulle , Lille 3)
      • Functional estimation in high dimensional data: Application to classification
    • 17:00. Camille Brunet (University Paris 10)
      • Discriminative clustering for high-dimensional data

Friday April 6, 2012

  • Session 3: Statistical learning and structured data.
    • 08:30. Brendan Murphy (University College Dublin)
      • Exploring Clustering Structure in Ranking Data
    • 09:30. Coffee break
    • 10:00. Mohamed Nadif (Université Paris 5)
      • Co-clustering under different approaches
    • 11:00. Pierre Latouche (University Paris 1 Panthéon-Sorbonne)
      • Complexity control in overlapping stochastic block models
  • Session 4: New and future problems in statistical learning.
    • 13:30. Pascal Massart (University Paris Sud)
      • Data-driven penalties: heuristics, results and thoughts...
    • 14:30. Coffee break
    • 15:00. Jérôme Saracco (University Bordeaux 1)
      • A sliced inverse regression approach for block-wise evolving data streams
    • 16:00. Gilles Blanchard (University Potsdam)
      • Transfer to an Unlabeled Task using kernel marginal predictors

Directions

The workshop will be held in IUT "A" , Boulevard Paul Langevin, 59653 Villeneuve d'Ascq Cedex. Directions, useful informations (in french) and maps can be found here.

Organization

The workshop is organized by:

with the support of: