A probabilistic incremental proximal gradient method

Abstract : In this paper, we propose a probabilistic optimization method, named probabilistic incremental proximal gradient (PIPG) method, by developing a probabilistic interpretation of the incremental proximal gradient algorithm. We explicitly model the update rules of the incremental proximal gradient method and develop a systematic approach to propagate the uncertainty of the solution estimate over iterations. The PIPG algorithm takes the form of Bayesian filtering updates for a state-space model constructed by using the cost function. Our framework makes it possible to utilize well-known exact or approximate Bayesian filters, such as Kalman or extended Kalman filters, to solve large-scale regularized optimization problems.
Type de document :
Pré-publication, Document de travail
5 pages. 2018
Liste complète des métadonnées

https://hal.archives-ouvertes.fr/hal-01946642
Contributeur : Emilie Chouzenoux <>
Soumis le : jeudi 6 décembre 2018 - 11:27:19
Dernière modification le : samedi 8 décembre 2018 - 01:40:27

Lien texte intégral

Identifiants

  • HAL Id : hal-01946642, version 1
  • ARXIV : 1812.01655

Citation

Ömer Deniz Akyildiz, Emilie Chouzenoux, Víctor Elvira, Joaquín Míguez. A probabilistic incremental proximal gradient method. 5 pages. 2018. 〈hal-01946642〉

Partager

Métriques

Consultations de la notice

46