Zobrazeno 1 - 10
of 15
pro vyhledávání: '"Mathieu Timothee"'
Publikováno v:
ESAIM: Proceedings and Surveys, Vol 74, Pp 119-136 (2023)
Some recent contributions to robust inference are presented. Firstly, the classical problem of robust M-estimation of a location parameter is revisited using an optimal transport approach - with specifically designed Wasserstein-type distances - that
Externí odkaz:
https://doaj.org/article/31095b126eca43378524c3facaeae5e0
Autor:
Degenne, Rémy, Mathieu, Timothée
We prove lower bounds on the error of any estimator for the mean of a real probability distribution under the knowledge that the distribution belongs to a given set. We apply these lower bounds both to parametric and nonparametric estimation. In the
Externí odkaz:
http://arxiv.org/abs/2403.01892
We investigate the regret-minimisation problem in a multi-armed bandit setting with arbitrary corruptions. Similar to the classical setup, the agent receives rewards generated independently from the distribution of the arm chosen at each time. Howeve
Externí odkaz:
http://arxiv.org/abs/2309.16563
Autor:
Mathieu, Timothée, Della Vecchia, Riccardo, Shilova, Alena, Centa, Matheus Medeiros, Kohler, Hector, Maillard, Odalric-Ambrym, Preux, Philippe
Recently, the scientific community has questioned the statistical reproducibility of many empirical results, especially in the field of machine learning. To solve this reproducibility crisis, we propose a theoretically sound methodology to compare th
Externí odkaz:
http://arxiv.org/abs/2306.10882
We study the corrupted bandit problem, i.e. a stochastic multi-armed bandit problem with $k$ unknown reward distributions, which are heavy-tailed and corrupted by a history-independent adversary or Nature. To be specific, the reward obtained by playi
Externí odkaz:
http://arxiv.org/abs/2203.03186
Autor:
Mathieu, Timothée
Publikováno v:
Electron. J. Statist. 16 (1) 3695 - 3750, 2022
We present a new finite-sample analysis of M-estimators of locations in $\mathbb{R}^d$ using the tool of the influence function. In particular, we show that the deviations of an M-estimator can be controlled thanks to its influence function (or its s
Externí odkaz:
http://arxiv.org/abs/2104.04416
Autor:
Minsker, Stanislav, Mathieu, Timothée
This paper investigates robust versions of the general empirical risk minimization algorithm, one of the core techniques underlying modern statistical methods. Success of the empirical risk minimization is based on the fact that for a "well-behaved"
Externí odkaz:
http://arxiv.org/abs/1910.07485
We present an extension of Vapnik's classical empirical risk minimizer (ERM) where the empirical risk is replaced by a median-of-means (MOM) estimator, the new estimators are called MOM minimizers. While ERM is sensitive to corruption of the dataset
Externí odkaz:
http://arxiv.org/abs/1808.03106
Mean embeddings provide an extremely flexible and powerful tool in machine learning and statistics to represent probability distributions and define a semi-metric (MMD, maximum mean discrepancy; also called N-distance or energy distance), with numero
Externí odkaz:
http://arxiv.org/abs/1802.04784
Publikováno v:
ESAIM: Proceedings and Surveys
Journées MAS de la SMAI, édition 2021.
ESAIM: Proceedings and Surveys, EDP Sciences, In press
ESAIM: Proceedings and Surveys, In press
Journées MAS de la SMAI, édition 2021.
ESAIM: Proceedings and Surveys, EDP Sciences, In press
ESAIM: Proceedings and Surveys, In press
Some recent contributions to robust inference are presented. Firstly, the classical problem of robust M-estimation of a location parameter is revisited using an optimal transport approach-with specifically designed Wasserstein-type distances-that red
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=dedup_wf_001::ddc68629d0f4e16842f07b0f1516c64f
https://hal.archives-ouvertes.fr/hal-03605702/file/Session_Robust_Learning_ESAIM_PROC.pdf
https://hal.archives-ouvertes.fr/hal-03605702/file/Session_Robust_Learning_ESAIM_PROC.pdf