Many modern applications rely on solving optimization problems (e.g., computational biology, mechanics, finance), establishing optimization methods as crucial tools in many scientific fields. Providing guarantees on the (hopefully good) behaviors of these methods is therefore of significant interest. A standard way of analyzing optimization algorithms consists in worst-case reasoning. That is, providing guarantees on the behavior of an algorithm (e.g. its convergence speed), that are independent of the function on which the algorithm is applied and true for every function in a particular class. This thesis aims at providing worst-case analyses of a few efficient first-order optimization methods. We start by the study of Anderson acceleratio...
main paper (9 pages) + appendix (21 pages)International audienceWe introduce a generic scheme for ac...
Several important problems in learning theory and data science involve high-dimensional optimization...
Minor modifications including acknowledgments and references. Code available at https://github.com/m...
Many modern applications rely on solving optimization problems (e.g., computational biology, mechani...
The goal of this work is to determine the performance of different first-order methods. To do it, we...
Optimization is an important discipline of applied mathematics with far-reaching applications. Optim...
We present a MATLAB toolbox that automatically computes tight worst-case performance guarantees for ...
We are interested in determining the worst performance exhibited by a given first-order optimization...
In many different fields such as optimization, the performance of a method is often characterized by...
We provide a framework for computing the exact worst-case performance of any algorithm belonging to ...
Plusieurs problèmes importants issus de l'apprentissage statistique et de la science des données imp...
This thesis mainly studies optimization algorithms. Programming problems arising in signal processin...
This thesis exposes contributions to the analysis of algorithms for noisy functions. It exposes conv...
This thesis aims at developing efficient optimization algorithms for solving large-scale machine lea...
This thesis aims at developing efficient algorithms for solving some fundamental engineering problem...
main paper (9 pages) + appendix (21 pages)International audienceWe introduce a generic scheme for ac...
Several important problems in learning theory and data science involve high-dimensional optimization...
Minor modifications including acknowledgments and references. Code available at https://github.com/m...
Many modern applications rely on solving optimization problems (e.g., computational biology, mechani...
The goal of this work is to determine the performance of different first-order methods. To do it, we...
Optimization is an important discipline of applied mathematics with far-reaching applications. Optim...
We present a MATLAB toolbox that automatically computes tight worst-case performance guarantees for ...
We are interested in determining the worst performance exhibited by a given first-order optimization...
In many different fields such as optimization, the performance of a method is often characterized by...
We provide a framework for computing the exact worst-case performance of any algorithm belonging to ...
Plusieurs problèmes importants issus de l'apprentissage statistique et de la science des données imp...
This thesis mainly studies optimization algorithms. Programming problems arising in signal processin...
This thesis exposes contributions to the analysis of algorithms for noisy functions. It exposes conv...
This thesis aims at developing efficient optimization algorithms for solving large-scale machine lea...
This thesis aims at developing efficient algorithms for solving some fundamental engineering problem...
main paper (9 pages) + appendix (21 pages)International audienceWe introduce a generic scheme for ac...
Several important problems in learning theory and data science involve high-dimensional optimization...
Minor modifications including acknowledgments and references. Code available at https://github.com/m...