AI Paper Atlas

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

FedReLa: Imbalanced Federated Learning via Re-Labeling

Guangzheng Hu, Patricia Menendez, Feng Liu, Mingming Gong, Guanghui Wang, Liuhua Peng

ICML 2026 · Federated learning and privacy › Personalized and heterogeneous FL · approach: Non-IID federated learning · 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 FedReLa one-shot re-labelingFedLOGE · CIFAR-10-LT IF=100 alpha=0.1overall accuracy: 45.08 -> 62.16Improved
add FedReLa one-shot re-labelingFedLOGE · CIFAR-100-LT IF=100 alpha=0.1overall accuracy: 29.21 -> 38.08Improved
add FedReLa one-shot re-labelingFedETF · CIFAR-10-LT IF=50 alpha=0.1overall accuracy: 47.96 -> 61.71Improved
add FedReLa to FedAvgFedAvg · CIFAR-10 step-wise IR=20 10% minorityminority accuracy: 17.25 -> 51.60Improved
add FedReLa to FedProxFedProx · CIFAR-10 step-wise IR=20 10% minorityminority accuracy: 15.60 -> 53.90Improved
add FedReLa to FedAvgFedAvg · CIFAR-10 step-wise IR=10 30% minoritymajority accuracy: 65.93 -> 62.14Hurt
add FedReLa to FedAvgFedAvg · CIFAR-10 step-wise IR=10 10% minoritymajority accuracy: 60.10 -> 59.84No clear effect
add FedReLa to FedYoYoFedYoYo · ImageNet-LT alpha=0.1overall accuracy: 38.15 -> 38.78Improved
add FedReLa to FedLoGeFedLoGe · ImageNet-LT alpha=0.1overall accuracy: 30.52 -> 31.70Improved
use z-score + tanh instead of min-max normalization for re-labeling probability (ablation)FedETF + FedReLa · CIFAR-100-LT IF=100tail accuracy: 21.46 -> 24.52Improved
use z-score + tanh instead of min-max normalization (ablation)FedETF + FedReLa · CIFAR-100-LT IF=100overall accuracy: 43.33 -> 44.54Improved
add FedReLa to FedETF as a baseline (ablation)FedETF · CIFAR-100-LT IF=100overall accuracy: 42.81 -> 43.33Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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