Incremental Without Replacement Sampling in Nonconvex Optimization - Argumentation, Décision, Raisonnement, Incertitude et Apprentissage Accéder directement au contenu
Pré-Publication, Document De Travail Année : 2021

Incremental Without Replacement Sampling in Nonconvex Optimization

Résumé

Minibatch decomposition methods for empirical risk minimization are commonly analysed in a stochastic approximation setting, also known as sampling with replacement. On the other hands modern implementations of such techniques are incremental: they rely on sampling without replacement, for which available analysis are much scarcer. We provide convergence guaranties for the latter variant by analysing a versatile incremental gradient scheme. For this scheme, we consider constant, decreasing or adaptive step sizes. In the smooth setting we obtain explicit complexity estimates in terms of epoch counter. In the nonsmooth setting we prove that the sequence is attracted by solutions of optimality conditions of the problem.
Fichier principal
Vignette du fichier
incrementalNonsmoothNonconvex.pdf (291.64 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-02896102 , version 1 (10-07-2020)
hal-02896102 , version 2 (19-04-2021)
hal-02896102 , version 3 (15-06-2021)
hal-02896102 , version 4 (26-12-2022)

Identifiants

Citer

Edouard Pauwels. Incremental Without Replacement Sampling in Nonconvex Optimization. 2021. ⟨hal-02896102v2⟩

Collections

SMS
181 Consultations
187 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More