Proximal Multitask Learning over Distributed Networks with Jointly Sparse Structure - ANR - Agence nationale de la recherche Accéder directement au contenu
Communication Dans Un Congrès Année : 2020

Proximal Multitask Learning over Distributed Networks with Jointly Sparse Structure

Danqi Jin
  • Fonction : Auteur
  • PersonId : 1110487
Jie Chen
Jingdong Chen
  • Fonction : Auteur
  • PersonId : 1110488
Cédric Richard

Résumé

Modeling relations between local optimum parameter vectors in multitask networks has attracted much attention over the last years. This work considers a distributed optimization problem for parameter vectors with a jointly sparse structure among nodes, that is, the parameter vectors share the same support set. By introducing an L∞,1-norm penalty at each node, and using a proximal gradient method to minimize the regularized cost, we devise a proximal multitask diffusion LMS algorithm which promotes the joint-sparsity to enhance the estimation performance. Analyses are provided to ensure the stability. Simulation results are presented to highlight the performance.
Fichier principal
Vignette du fichier
jin2019proximal.pdf (634.38 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-03347335 , version 1 (17-09-2021)

Identifiants

Citer

Danqi Jin, Jie Chen, Jingdong Chen, Cédric Richard. Proximal Multitask Learning over Distributed Networks with Jointly Sparse Structure. ICASSP 2020 - 2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), May 2020, Barcelona, France. pp.5900-5904, ⟨10.1109/ICASSP40776.2020.9053579⟩. ⟨hal-03347335⟩
16 Consultations
32 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More