Passer à la navigation principale Passer à la recherche Passer au contenu principal

Optimal algorithms for smooth and strongly convex distributed optimization in networks

  • Kevin Seaman
  • , Francis Bach
  • , Sebastien Bubeck
  • , Yin Tat Lee
  • , Laurent Massoulie
  • INRIA
  • PSL research University & IPSL

Résultats de recherche: Le chapitre dans un livre, un rapport, une anthologie ou une collectionContribution à une conférenceRevue par des pairs

Résumé

Theory group, Microsoft Research, Redmond, United Stales, In this paper, wc determine the optimal convergence rates for strongly convex and smooth distributed optimization in two settings: centralized and decentralized communications over a network. For centralized (i.e. master/slave) algorithms, we show that distributing Nesterov's accelerated gradient descent is optimal and achieves a precision e > 0 in time 0(KJ(1 + Ar)ln(l/e)), where Kg is the condition number of the (global) function to optimize, A is the diameter of the network, and r (resp. 1) is the time needed to communicate values between two neighbors (resp. perform local computations). For decentralized algorithms based on gossip, we provide the first optimal algorithm, called the multi-step dual accelerated (MSDA) method, that achieves a precision e > 0 in time 0 0 7 ( 1 + where KJ is the condition number of the local functions and 7 is the (normalized) eigengap of the gossip matrix used for communication between nodes. We then verify the efficiency of MSDA against state-of-theart methods for two problems: least-squares regression and classification by logistic regression.

langue originaleAnglais
titre34th International Conference on Machine Learning, ICML 2017
EditeurInternational Machine Learning Society (IMLS)
Pages4630-4642
Nombre de pages13
ISBN (Electronique)9781510855144
étatPublié - 1 janv. 2017
Modification externeOui
Evénement34th International Conference on Machine Learning, ICML 2017 - Sydney, Australie
Durée: 6 août 201711 août 2017

Série de publications

Nom34th International Conference on Machine Learning, ICML 2017
Volume6

Une conférence

Une conférence34th International Conference on Machine Learning, ICML 2017
Pays/TerritoireAustralie
La villeSydney
période6/08/1711/08/17

Empreinte digitale

Examiner les sujets de recherche de « Optimal algorithms for smooth and strongly convex distributed optimization in networks ». Ensemble, ils forment une empreinte digitale unique.

Contient cette citation