Laboratoire de Mathématiques Raphaël Salem

UMR 6085 CNRS - Université de Rouen Normandie

In the context of large samples, a small number of individuals can alter
basic statistical indicators such as the mean. It is difficult to automatically detect these
atypical individuals, and an alternative strategy is to use robust approaches. This work
focuses on the estimation of the geometric median of a random variable, which is a robust
indicator of location. In order to deal with large samples or sequentially arriving data,
online stochastic Newton algorithms for geometric median estimation are introduced and
their rates of convergence are given.