AI Paper Atlas

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

FedRGL: Robust Federated Graph Learning under Label Noise

De Li, Zhou Tan, Qiyu Li, Zeming Gan, Tiange Xia, Chunpei Li, Xianxian LI

ICML 2026 · Federated learning and privacy › Personalized and heterogeneous FL · approach: Federated semi-supervised and graph learning

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 FedRGL on top of FedAvg under Uniform noiseFedAvg · Cora (5 clients), Uniform noise 0.3accuracy: 48.19 -> 78.75Improved
add FedRGL on top of FedAvg under Pair noiseFedAvg · Cora (5 clients), Pair noise 0.3accuracy: 55.80 -> 75.14Improved
add FedRGL on top of FedAvg under Uniform noiseFedAvg · Photo (20 clients), Uniform noise 0.3accuracy: 72.55 -> 84.06Improved
use FedRGL instead of FedAvg with clean labelsFedAvg · PubMed (10 clients), clean labelsaccuracy: 85.86 -> 85.33No clear effect
use FedRGL instead of CRGNN with clean labelsCRGNN · Cora (5 clients), clean labelsaccuracy: 83.09 -> 82.12Hurt
use FedRGL instead of CRGNN under Uniform noiseCRGNN · CS (10 clients), Uniform noise 0.3accuracy: 81.30 -> 87.79Improved
add FedRGL as the number of noisy clients rises to 100%FedAvg · Photo (20 clients), Pair noise 0.4, 100% noisy clientsaccuracy: 53.4 -> 73.4Improved
use FedRGL with few noisy clients (20%)FedAvg · Photo (20 clients), Pair noise 0.4, 20% noisy clientsaccuracy: 85.7 -> 87.5Improved
apply FedRGL on large-scale graph with 20 clientsFedAvg · ogbn-arxiv (20 clients), Uniform noise 0.4accuracy: 51.83 -> 56.61Improved
apply FedRGL on large-scale graph with 30 clients, Pair noiseFedAvg · ogbn-arxiv (30 clients), Pair noise 0.4accuracy: 45.33 -> 50.83Improved
add the global model view to noisy node filtering (ablation)FedRGL · Cora and CS, noise 0.3accuracyImproved
add the local structure view to noisy node filtering (ablation)FedRGL · Cora and CS, noise 0.3accuracyImproved

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