v2.12.0 (512)

Cours scientifiques - APM_4OPT2_TA : Optimisation différentiable 2

Descriptif

This second part follows OPT201. The course will be held in English.

The OPT202 course covers the algorithmic aspects of optimization, and in particular first-order and second-order algorithms to solve optimization problems as well as their theoretical complexities and guarantees. 

Objectifs pédagogiques

To be able :
- to set up algorithms to solve optimization problems
- to be able to prove their theoretical properties. 

21 heures en présentiel

Diplôme(s) concerné(s)

Parcours de rattachement

Format des notes

Numérique sur 20

Littérale/grade européen

Programme détaillé

1. CM: Into and unconstraint optimization I:
        complexity classes, gradient and newton's method for non-convex problems
 2.   CM: unconstraint optimization II (convex):
        Gradient, Nesterov's method, complexity classes, subgradient.

3.   CM: constrained optimization I (first-order):
        proximal gradient, dual ascent, primal-dual.

4. CM: constrained optimization II (second-order):
        self-concordand barriers, interiorior-point method

Mots clés

Optimisation, Conditions d'optimalité, Méthodes Numériques, Algorithmes, Gradient, Newton, Quasi-Newton, Méthode de points intérieurs, Dualité
Veuillez patienter