AI Paper Atlas

Research map / Federated learning and privacy / Personalized and heterogeneous FL

Federated Learning for Non-IID Data via Unified Feature Learning and Optimization Objective Alignment

Lin Zhang, Yong Luo, Yan Bai, Bo Du, Ling-Yu Duan

ICCV 2021 · Federated learning and privacy › Personalized and heterogeneous FL · approach: Federated learning with pretrained models · 100 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 uniform adversarial (UAD) loss (ablation)FedAVG · CIFAR-10accuracy: 59.95 -> 65.29Improved
add group consensus (CGR) loss (ablation)FedAVG · CIFAR-10accuracy: 59.95 -> 65.10Improved
add global consensus (CGL) loss (ablation)FedAVG · CIFAR-10accuracy: 59.95 -> 65.01Improved
add CGR loss on top of UAD (ablation)UAD · CIFAR-10accuracy: 65.29 -> 66.29Improved
add CGL loss on top of UAD (ablation)UAD · CIFAR-10accuracy: 65.29 -> 66.30Improved
add both consensus losses on top of UAD (ablation)UAD · CIFAR-10accuracy: 65.29 -> 67.80Improved
add UAD loss on top of CGR loss (ablation)CGR · CIFAR-10accuracy: 65.10 -> 66.29Improved
add dynamic sampling strategy (ablation)All three losses · CIFAR-10accuracy: 67.80 -> 67.45Hurt
add dynamic sampling strategy (ablation)All three losses · CIFAR-10Pstd (fairness, lower better): 7.42 -> 3.54Improved
add dynamic sampling strategy (ablation)CGL only · CIFAR-10accuracy: 65.01 -> 63.07Hurt
add UAD loss (ablation)FedAVG · CIFAR-10Pstd (fairness, lower better): 9.23 -> 4.21Improved
add CGR loss on top of UAD+CGL (ablation)UAD + CGL · CIFAR-10Pstd (fairness, lower better): 4.21 -> 7.07Hurt

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