AI Paper Atlas

Research map / Federated learning and privacy / Communication-efficient FL

A Computation and Communication Efficient Method for Distributed Nonconvex Problems in the Partial Participation Setting

Alexander Tyurin, Peter Richtárik

NeurIPS 2023 · Federated learning and privacy › Communication-efficient FL · approach: Federated optimization algorithms · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
reduce participation to s=10 of n=100 nodes (probability 1/10) (ablation)DASHA-PP · real-sim, finite-sum and stochasticconvergence speed (gradient norm vs communication rounds): approximately x10 slowerHurt
reduce participation to s=1 of n=100 nodes (probability 1/100) (ablation)DASHA-PP · real-sim, finite-sum and stochasticconvergence speed (gradient norm vs communication rounds): approximately x100 slowerHurt
use DASHA-PP instead of MARINA under partial participation with RandKDASHA-PP · real-sim and MNIST, finite-sumconvergence (gradient norm vs rounds)Improved
use DASHA-PP instead of FRECON in finite-sum settingDASHA-PP · real-sim and MNIST, finite-sumfinal accuracy of solutionImproved
use DASHA-PP instead of MARINA and FRECON in stochastic settingDASHA-PP · real-sim and MNIST, stochasticconvergence (gradient norm)Improved
increase participation probability toward 1 (ablation)DASHA-PP vs DASHA · real-simconvergence rate gap to DASHAImproved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy