Bayesian learning theory for complex data modelling

Recent successes of machine learning in the big data era are radically widening the gap between statistical theory and practice for the analysis of complex data structures, such as social networks data, multi-layer and temporal networks, images flows, text document flows, just to mention a few. Off-the-shelf statistical and machine learning methods are commonly used by practitioners while a proper understanding of the mathematical properties of such algorithms is generally lacking. The Bayesian paradigm somehow trikes a balance between theory and practice by enabling sound mathematical guarantees, efficient implementation and model selection criteria. The workshop targets those researchers interested in the recent developments in Bayesian learning theory along different directions including extreme value theory, PAC-Bayesian methods, Bayesian Nonparametrics, regularised methods, etc.

The workshop will give the opportunity to learn from and interact with highly qualified senior researchers in probability, theoretical and applied statistics, with a particular focus on Bayesian methods.

Link to 1st IFSS edition.

Keynote speakers

• Nicolas Chopin, Ensae, France (cancelled)
• Arnoldo Frigessi, University of Oslo, Norway
• Antonio Lijoi, University Bocconi, Milan, Italy
• Sonia Petrone, University Bocconi, Milan, Italy

Confirmed speakers

• Simon Barthelmé, Gipsa-lab, Grenoble, France
• Annalisa Cadonna, WU Vienna, Austria
• Benjamin Guedj, Inria Lille - Nord Europe, France
• Alessandra Guglielmi, Politecnico di Milano, Italy
• Bernardo Nipoti, Trinity College Dublin, Ireland
• Elodie Vernet, Ecole polytechnique, France

Registration is free but mandatory

Please register by the 25th of August 2018.