AI Paper Atlas

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

Prior Refinement Is Better: Diffusion-Driven Graph Harmonization for Federated Graph Learning

Shuman Zhuang, Zhihao Wu, Wei Huang, Luojun Lin, Jia-Li Yin, Lele Fu, Hong-Ning Dai

AAAI 2026 · Federated learning and privacy › Personalized and heterogeneous FL · approach: Federated semi-supervised and graph learning · 1 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 FedGH prior refinement to FedTADFedTAD · Cora (20 clients)accuracy: 63.74 -> 67.28Improved
add FedGH prior refinement to FedTADFedTAD · PubMed (20 clients)accuracy: 81.19 -> 83.73Improved
add FedGH prior refinement to FedTADFedTAD · Photo (10 clients)accuracy: 87.76 -> 87.92No clear effect
add FedGH prior refinement to FedTADFedTAD · Photo (20 clients)accuracy: 84.94 -> 86.29Improved
add FedGH prior refinement to FedTADFedTAD · Cora (20 clients)F1-score: 60.34 -> 63.90Improved
add FedGH prior refinement to FedTADFedTAD · PubMed (10 clients)F1-score: 80.26 -> 80.47No clear effect
use FedGH instead of FedAvg aloneFedAvg · Cora (20 clients)accuracy: 62.4 -> 70.7Improved
use FedGH instead of FedATHFedATH · ogbn-arxiv (20 clients)accuracy: 38.9 -> 38.1Hurt
add the KL divergence term to the VGAE loss (ablation)FedGH · Cora and PubMed (10, 15, 20 clients)accuracyImproved
add the conditional vector to the diffusion model (ablation)FedGH · Cora and PubMed (10, 15, 20 clients)accuracyImproved
add contrastive learning on pseudo-graphs (ablation)FedGH · Cora and PubMed (10, 15, 20 clients)accuracyImproved
vary the KL weight gammaFedGH · Cora and PubMed (10, 15, 20 clients)accuracyNo clear effect

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