Le but de ces journées est de donner une vue d'ensemble des développements scientifiques récents en statistique et de promouvoir les échanges entre étudiants diplômés, chercheurs confirmés.
-
Mini-cours
2 x 1h30Claire Boyer - LPSM, Sorbonne Université - Jeudi 20/06 à 16h30 et Vendredi 21/06 à 11h
A primer on diffusion-based generative modelsEmmanuel Rachelson - ISAE-Supaéro - Mercredi 19/06 à 11h et Jeudi 20/06 à 11h
Introduction à l'apprentissage par renforcement -
Exposés longs
45 min + 15 min questionsNicolas Chopin - ENSAE, Institut Polytechnique de Paris - TBA
Unbiased estimation of smooth functions, Applications in statistic and machine learningMaud Delattre - INRAE, Unité MaIAGE - TBA
A new preconditioned stochastic gradient algorithm for estimation in latent variable models.Edith Gabriel - INRAE, Unité BioSP - TBA
Integrated Spatial Surveillance for Risk Assessment Using Point Processes and Machine Learning: Application to Plant HealthSébastien Gerchinovitz - IRT St Exupéry - TBA
Guaranteed prediction sets via concentration inequalitiesAnaïs Rouanet - ISPED, Université de Bordeaux - TBA
Nonparametric Bayesian mixture models for identifying clusters from longitudinal and cross-sectional dataFrédéric Richard - I2M, Université Aix-Marseille - TBA
Inference techniques for the analysis of Brownian image texturesMathieu Serrurier - IRIT, Université Paul-Sabatier - TBA
Building explainable and robust neural networks by using Lipschitz constraints and Optimal TransportThe lack of robustness and explainability in neural networks is directly linked to the arbitrarily high Lipschitz constant of deep models. Although constraining the Lipschitz constant has been shown to improve these properties, it can make it challenging to learn with classical loss functions. In this presentation, we explain how to control this constant, and demonstrate that training such networks requires defining specific loss functions and optimization processes. To this end, we propose a loss function based on optimal transport that not only certifies robustness but also converts adversarial examples into provable counterfactual examples.
TBA
-
Exposés courts
20 min + 10 min questionsMarie Chion - MRC Biostatistics Unit, University of Cambridge - TBA
-
Session poster
TBA
-
Pré-journée pour les étudiant.es
20 min + 5 min questionsHanna Bacave - INRAE, MIAT - TBA
TBANicolas Enjalbert Courrech - IMT, CNRS - TBA
Comparaison de méthodes d'inférence post-clusteringLilit Hovsepyan - Université Le Mans, Inrae Toulouse - TBA
Fast inference in copula regression models with categorical explanatory variables using one-step proceduresSophia Yazzourh - IMT, Université Paul Sabatier - TBA
TBA
