AI Paper Atlas

Research map / Graph learning / Graph self-supervised learning

Heterogeneity-Aware Knowledge Sharing for Graph Federated Learning

Wentao Yu, Sheng Wan, Shuo Chen, Bo Han, Chen Gong

ICML 2026 · Graph learning › Graph self-supervised learning · approach: Heterophily and imbalance on graphs · 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 semantic knowledge sharing (ablation)FedSSA · Cora and TolokersaccuracyImproved
add structural knowledge sharing (ablation)FedSSA · Cora and TolokersaccuracyImproved
add both semantic and structural knowledge sharing (ablation)FedSSA · Coraaccuracy: more than 5%Improved
use FedSSA instead of second-best FedIIHFedSSA · heterophilic datasetsaverage accuracy: 2.82%Improved
use FedSSA instead of FedIIH on homophilic averageFedSSA · six homophilic datasetsaverage accuracy: 83.10 -> 83.53Improved
use FedSSA instead of FedIIH on heterophilic averageFedSSA · five heterophilic datasetsaverage accuracy/AUC: 64.52 -> 67.34Improved
use FedSSA instead of FedIIHFedSSA · Minesweeper (10 clients)AUC: 73.23 -> 82.16Improved
add semantic and structural knowledge sharing for faster convergenceFedSSA · non-overlapping, 10 clientscommunication rounds to converge: 90 -> 60Improved
vary K_node, K_struct, lambda_1, lambda_2 (ablation)FedSSA · CoraaccuracyNo clear effect
use FedSSA instead of Local training on Amazon-ratingsFedSSA · Amazon-ratings (10 clients)accuracy: 45.89 -> 45.11No 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