Best poster award for Nora Lüthen at MASCOT-NUM 2021
Nora received the award for her poster on “Sparse regression-based Poincaré expansions for global sensitivity analysis”, which she presented at this year’s virtual edition of the annual MASCOT-NUM 2021 meeting.
Several members of the Chair participated in this year’s external pageMASCOT-NUM 2021 conferencecall_made, the annual meeting of the French research association external pageGdR MASCOT-NUMcall_made (Méthodes d’Analyse Stochastique pour les COdes et Traitements Numériques) specialized on stochastic methods for the analysis of computational models.
Nora presented the results of a fruitful collaboration between the Chair and external pageFabrice Gamboacall_made (Institut de Mathématiques de Toulouse), external pageBertrand Ioosscall_made (EDF R&D), and external pageOlivier Roustantcall_made (INSA Toulouse). The investigated Poincaré chaos expansions are similar in spirit to the established surrogate modelling method of Polynomial chaos expansions, but in addition naturally allow incorporating derivative information into the surrogate modelling process, which appears to be especially promising for Sobol’ sensitivity analysis. A joint paper is in preparation.
The poster award is endowed with 500€. Another best poster prize was awarded to Athénaïs Gauthier (Bern University) for her work on “Sample-based estimation of probability density fields: a spatial extension of the logistic Gaussian process”. Clément Bénard (Safrantech/Sorbonne University) received the best presentation award for his presentation on “Random forests: a sensitivity analysis perspective”.