Augmented cumulative distribution networks for multivariate extreme value modelling

Gildas Mazo 1 Florence Forbes 1 Stephane Girard 1
1 MISTIS - Modelling and Inference of Complex and Structured Stochastic Systems
Inria Grenoble - Rhône-Alpes, LJK - Laboratoire Jean Kuntzmann, INPG - Institut National Polytechnique de Grenoble
Abstract : Max-stable distribution functions are theoretically grounded models for modelling multivariate extreme values. However they suffer from some striking limitations when applied to real data analysis due to the intractability of the likelihood when the number of variables becomes high. Cumulative Distribution Networks (CDN's) have been introduced recently in the machine learning community and allow the construction of max-stable distribution functions for which the density can be computed. Unfortunately, we show in this work that the dependence structure expected in the data may not be accurately reflected by max-stable CDN's. To face this limitation, we therefore propose to augment max-stable CDN's with the more standard Gumbel max-stable distribution function in order to enrich the dependence structure.
Type de document :
Communication dans un congrès
ERCIM 2012 - 5th International Conference of the ERCIM WG on Computing and Statistics, Dec 2012, Oviedo, Spain. 2012
Liste complète des métadonnées

https://hal.archives-ouvertes.fr/hal-00803444
Contributeur : Stephane Girard <>
Soumis le : vendredi 22 mars 2013 - 08:59:52
Dernière modification le : mercredi 11 avril 2018 - 01:57:53

Identifiants

  • HAL Id : hal-00803444, version 1

Collections

Citation

Gildas Mazo, Florence Forbes, Stephane Girard. Augmented cumulative distribution networks for multivariate extreme value modelling. ERCIM 2012 - 5th International Conference of the ERCIM WG on Computing and Statistics, Dec 2012, Oviedo, Spain. 2012. 〈hal-00803444〉

Partager

Métriques

Consultations de la notice

298