A principled framework for the design and analysis of token algorithms - Archive ouverte HAL Access content directly
Conference Papers Year : 2023

A principled framework for the design and analysis of token algorithms

Abstract

We consider a decentralized optimization problem, in which n nodes collaborate to optimize a global objective function using local communications only. While many decentralized algorithms focus on gossip communications (pairwise averaging), we consider a different scheme, in which a "token" that contains the current estimate of the model performs a random walk over the network, and updates its model using the local model of the node it is at. Indeed, token algorithms generally benefit from improved communication efficiency and privacy guarantees. We frame the token algorithm as a randomized gossip algorithm on a conceptual graph, which allows us to prove a series of convergence results for variance-reduced and accelerated token algorithms for the complete graph. We also extend these results to the case of multiple tokens by extending the conceptual graph, and to general graphs by tweaking the communication procedure. The reduction from token to well-studied gossip algorithms leads to tight rates for many token algorithms, and we illustrate their performance empirically.
Fichier principal
Vignette du fichier
2205.15015.pdf (833.25 Ko) Télécharger le fichier
Origin : Files produced by the author(s)

Dates and versions

hal-04107248 , version 1 (26-05-2023)

Identifiers

Cite

Hadrien Hendrikx. A principled framework for the design and analysis of token algorithms. AISTATS 2023 - 26th International Conference on Artificial Intelligence and Statistics, Apr 2023, Valencia, Spain. pp.470-489. ⟨hal-04107248⟩
41 View
28 Download

Altmetric

Share

Gmail Facebook X LinkedIn More