AI Paper Atlas

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

Federated Learning with Client Subsampling, Data Heterogeneity, and Unbounded Smoothness: A New Algorithm and Lower Bounds

Michael Crawshaw, Yajie Bao, Mingrui Liu

NeurIPS 2023 · Federated learning and privacy › Communication-efficient FL · approach: Client participation analysis · 3 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
use EPISODE++ instead of CELGCfederated RNN · SNLItest accuracy: 77.45 -> 83.89Improved
use EPISODE++ instead of Clipped Minibatch SGDfederated RNN · SNLItest accuracy: 72.74 -> 83.89Improved
use EPISODE++ instead of NaiveParallelClip (with less communication)federated RNN · SNLItest accuracy: 81.74 -> 83.89Improved
use EPISODE++ instead of CELGCfederated RNN · Sentiment140test accuracy: 76.36 -> 77.97Improved
use EPISODE++ instead of NaiveParallelClipfederated RNN · Sentiment140train loss: 0.365 -> 0.258Improved
replace episodic clipping with per-step clipping (SCAFFOLDClip) (ablation)EPISODE++ · SNLItest accuracy: 83.89 -> 81.11Hurt
replace episodic clipping with per-step clipping (SCAFFOLDClip) (ablation)EPISODE++ · Sentiment140test accuracy: 77.97 -> 77.38Hurt
ignore information from previous rounds (EPISODE) (ablation)EPISODE++ · SNLItest accuracy: 83.89 -> 80.24Hurt
ignore information from previous rounds (EPISODE) (ablation)EPISODE++ · Sentiment140train loss: 0.258 -> 0.466Hurt
decrease number of participating clients SEPISODE++ · SNLI/Sentiment140test accuracyNo clear effect
decrease number of participating clients SClipped minibatch SGD, NaiveParallelClip, EPISODE · SNLI/Sentiment140test accuracyHurt
decrease data similarity sCELGC, Clipped minibatch SGD, NaiveParallelClip · SNLI/Sentiment140test accuracyHurt

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