AI Paper Atlas

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

Communication-Efficient Federated Bilevel Optimization with Global and Local Lower Level Problems

Junyi Li, Feihu Huang, Heng Huang

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
Add more clients sampled per epochFedBiOAcc · Federated Data Cleaning (MNIST, 10 clients)validation error vs communication roundsImproved
Increase number of local steps I under larger label noiseFedBiOAcc · Federated Data Cleaning (MNIST)validation error vs communication roundsImproved
Add momentum-based variance reduction (FedBiOAcc vs FedBiO) (ablation)FedBiO · Federated Data Cleaning and Hyper-representationvalidation error vs communication roundsImproved
Add multiple local steps (I=5) instead of one-step distributed version (ablation)FedBiOAcc-Local · Omniglot and MiniImageNet N-way-K-shotvalidation error vs communication roundsImproved
Compare FedBiOAcc with FedAvg trained on noisy data at high noise rho=0.95FedBiOAcc · Federated Data Cleaning (MNIST)validation errorImproved
Compare FedBiOAcc against FedNest, CommFedBiO, AggITD, FedMBO, Local-BSGVR across noise levelsFedBiOAcc · Federated Data Cleaning, rho 0.1 to 0.95validation error vs communication roundsImproved

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