

From robust tests to robust Bayes-like posterior distribution
By Yannick Baraud


Statistical learning in biological neural networks
By Johannes Schmidt-Hieber
Appears in collection : Extreme value theory and laws of rare events / Théorie des valeurs extrêmes et lois des évènements rares
Analysing heavy rainfall in France is complex due to the high number of weather stations and the complexity of weather system patterns over the French territory. This leads to computational issues and classical Extreme Value Theory (ETV) cannot be directly applied. To bypass the computational hurdles, we perform a dimension reduction approach, based on EVT concepts, to create independent regional clusters. Within each cluster, we propose a nonparametric approach for estimating the maxima dependence function. This is a joint work with S. A. Padoan, G. Marcon, E. Bernard, M. Vrac, O. Mestre.