In statistics and learning theory, it is common to assume that samples are independently and identically distributed according to a reference probability distribution. A more realistic approach could be to relax this assumption by allowing a fraction of samples to not necessarily follow the reference distribution. These disobeying samples, called outliers, may drastically skew the classical estimators. In this work, we aim to estimate the mean of reference distributions by estimators robust to outliers. We are interested in the non-asymptotic behavior of the estimators. In the first stage, we describe various contamination models which determine the nature of the outliers among our observations. Then, we consider the problem of estimating t...
We study the fundamental problem of learning the parameters of a high-dimensional Gaussian in the pr...
In this thesis, we are interested in estimating the mean of heavy-tailed random variables. We focus ...
We introduce a criterion, resilience, which allows properties of a dataset (such as its mean or best...
In statistics and learning theory, it is common to assume that samples are independently and identic...
In statistics and learning theory, it is common to assume that samples are independently and identic...
In statistics and learning theory, it is common to assume that samples are independently and identic...
In statistics and learning theory, it is common to assume that samples are independently and identic...
In statistics and learning theory, it is common to assume that samples are independently and identic...
In statistics and learning theory, it is common to assume that samples are independently and identic...
En statistique et en théorie de l'apprentissage statistique, on suppose souvent que les échantillons...
The problem of fitting a model to noisy data is fundamental to statistics and machine learning. In t...
The problem of fitting a model to noisy data is fundamental to statistics and machine learning. In t...
Many datasets are collected automatically, and are thus easily contaminated by outliers. In order to...
The design of statistical estimators robust to outliers has been a mainstay of statistical research ...
Presented on October 31, 2016 at 11:00 a.m. in the Klaus Advanced Computing Building, Room 1116EAnku...
We study the fundamental problem of learning the parameters of a high-dimensional Gaussian in the pr...
In this thesis, we are interested in estimating the mean of heavy-tailed random variables. We focus ...
We introduce a criterion, resilience, which allows properties of a dataset (such as its mean or best...
In statistics and learning theory, it is common to assume that samples are independently and identic...
In statistics and learning theory, it is common to assume that samples are independently and identic...
In statistics and learning theory, it is common to assume that samples are independently and identic...
In statistics and learning theory, it is common to assume that samples are independently and identic...
In statistics and learning theory, it is common to assume that samples are independently and identic...
In statistics and learning theory, it is common to assume that samples are independently and identic...
En statistique et en théorie de l'apprentissage statistique, on suppose souvent que les échantillons...
The problem of fitting a model to noisy data is fundamental to statistics and machine learning. In t...
The problem of fitting a model to noisy data is fundamental to statistics and machine learning. In t...
Many datasets are collected automatically, and are thus easily contaminated by outliers. In order to...
The design of statistical estimators robust to outliers has been a mainstay of statistical research ...
Presented on October 31, 2016 at 11:00 a.m. in the Klaus Advanced Computing Building, Room 1116EAnku...
We study the fundamental problem of learning the parameters of a high-dimensional Gaussian in the pr...
In this thesis, we are interested in estimating the mean of heavy-tailed random variables. We focus ...
We introduce a criterion, resilience, which allows properties of a dataset (such as its mean or best...