AI Paper Atlas

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

Federated Graph-Level Clustering Network with Dual Knowledge Separation

Xiaobao Wang, Renda Han, Ronghao Fu, Di Jin

ICLR 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 client-side subgraph separation and two-stage k-means (with FedAvg server) (ablation)FGCN-DKS · SM non-IIDACC (%): 61.7 -> 68.1Improved
add client-side subgraph separation and two-stage k-means (with FedAvg server) (ablation)FGCN-DKS · CV non-IIDACC (%): 29.5 -> 37.7Improved
add CKSS server aggregation without local separation (ablation)FGCN-DKS · SM non-IIDACC (%): 61.7 -> 64.6Improved
add CKSS server aggregation without local separation (ablation)FGCN-DKS · CV non-IIDACC (%): 29.5 -> 32.4Improved
add CKSS server aggregation on top of local separation (ablation)FGCN-DKS · SM non-IIDACC (%): 68.1 -> 79.2Improved
add CKSS server aggregation on top of local separation (ablation)FGCN-DKS · SM-BIO-SY non-IIDACC (%): 67.2 -> 73.6Improved
add local separation on top of CKSS (ablation)FGCN-DKS · SM-BIO non-IIDACC (%): 61.6 -> 73.8Improved
full FGCN-DKS vs prior best FedPKAFGCN-DKS · SM non-IIDACC (%): 77.0 -> 79.2Improved
full FGCN-DKS vs prior FedGCNFGCN-DKS · unequal cluster countsACC (%): 54.5 -> 62.1Improved
increase alpha:beta loss weight ratio above 1:1 (ablation)FGCN-DKS · four non-IID settingsACC, NMI, ARI, F1Mixed
use FGCN-DKS instead of FedGCN in communication cost (1 client, CV)FGCN-DKS · CV non-IIDcommunication time (s): 15.9 -> 16.8Hurt

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