26 janvier 2018: 1 événement

  • Séminaire étudiants

    Vendredi 26 janvier 14:30-15:30 - Antoine Godichon - INSA Rouen

    Séminaire étudiants : Clustering compositional data and applications

    Résumé : Although there is no shortage of clustering algorithms proposed in the literature, the question of the most relevant strategy for clustering compositional data (i.e., data whose rows belong to the simplex) remains largely unexplored. This work is motivated by the analysis of two applications, both focused on the categorization of compositional profiles : (1) identifying groups of co-expressed genes from high-throughput RNA sequencing data, in which a given gene may be completely silent in one or more experimental conditions ; and (2) finding patterns in the usage of stations over the course of one week in the Velib’ bicycle sharing system in Paris, France. For both of these applications, we make use of appropriately chosen data transformations, including the Centered Log Ratio and a novel extension called the Log Centered Log Ratio, in conjunction with the $K$-means algorithm.

    Lieu : A318

    En savoir plus : Séminaire étudiants

26 janvier 2018: 1 événement

  • Séminaire Mathématiques pour l’entreprise

    Vendredi 26 janvier 13:10-14:00 - Mathieu Ribatet - Université de Montpellier

    Séminaire Mathématiques pour l’entreprise : A journey along the sample path of a max-stable process

    Résumé : Max-stable processes play a major role in the areal modeling of spatial extremes, and more precisely pointwise maxima. This talk will cover some probabilistic as well as statistical aspects related to these processes. More precisely, we will review some (more or less) recent results about the hidden structure of max-stable processes through the notion of spectral characterization, extremal and sub-extremal functions as well as the hitting scenario. We will then see how one can achieve conditional simulations from max-stable processes using this hidden structure. Next a novel (spatial) dependence summary measure for spatial extremes, namely the extremal concurrence probability, will be introduced and the strongly connected notion of extremal concurrence cell. Finally, if we have enough time, we will see how full likelihood inference for max-stable processes is actually possible—though highly CPU demanding.

    Lieu : Salle René Baire

    En savoir plus : Séminaire Mathématiques pour l’entreprise