Dynamic Backup Workers for Parallel Machine Learning - Archive ouverte HAL Access content directly
Preprints, Working Papers, ... Year :

Dynamic Backup Workers for Parallel Machine Learning

(1) , (1) , (1)
1
Chuan Xu
Giovanni Neglia

Abstract

The most popular framework for distributed training of machine learning models is the (synchronous) parameter server (PS). This paradigm consists of n workers, which iteratively compute updates of the model parameters, and a stateful PS, which waits and aggregates all updates to generate a new estimate of model parameters and sends it back to the workers for a new iteration. Transient computation slowdowns or transmission delays can intolerably lengthen the time of each iteration. An efficient way to mitigate this problem is to let the PS wait only for the fastest n − b updates, before generating the new parameters. The slowest b workers are called backup workers. The correct choice of the number b of backup workers depends on the cluster configuration and workload, but also (as we show in this paper) on the hyper-parameters of the learning algorithm and the current stage of the training. We propose DBW, an algorithm that dynamically decides the number of backup workers during the training process to maximize the convergence speed at each iteration. Our experiments show that DBW 1) removes the necessity to tune b by preliminary time-consuming experiments, and 2) makes the training up to a factor 3 faster than the optimal static configuration.
Fichier principal
Vignette du fichier
revised_com_net (2).pdf (1.84 Mo) Télécharger le fichier
Origin : Files produced by the author(s)

Dates and versions

hal-03044199 , version 1 (07-12-2020)

Identifiers

  • HAL Id : hal-03044199 , version 1

Cite

Chuan Xu, Giovanni Neglia, Nicola Sebastianelli. Dynamic Backup Workers for Parallel Machine Learning. 2020. ⟨hal-03044199⟩
27 View
105 Download

Share

Gmail Facebook Twitter LinkedIn More