Pré-Publication, Document De Travail Année : 2023

One-step corrected projected stochastic gradient descent for statistical estimation

Résumé

A generic, fast and asymptotically efficient method for parametric estimation is described. It is based on the projected stochastic gradient descent on the log-likelihood function corrected by a single step of the Fisher scoring algorithm. We show theoretically and by simulations that it is an interesting alternative to the usual stochastic gradient descent with averaging or the adaptative stochastic gradient descent.

Fichier principal
Vignette du fichier
BE24.pdf (318.12 Ko) Télécharger le fichier
Origine Fichiers produits par l'(les) auteur(s)
Licence

Dates et versions

hal-04122876 , version 1 (08-04-2024)

Licence

Identifiants

  • HAL Id : hal-04122876 , version 1

Citer

Alexandre Brouste, Youssef Esstafa. One-step corrected projected stochastic gradient descent for statistical estimation. 2024. ⟨hal-04122876⟩
509 Consultations
357 Téléchargements

Partager

  • More