AI Paper Atlas

Research map / Graph learning / GNN architectures

Federated Graph Classification over Non-IID Graphs

Han Xie, Jing Ma, Li Xiong, Carl Yang

NeurIPS 2021 · Graph learning › GNN architectures · approach: Simple and scalable GNNs · 62 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
cluster clients by gradients (GCFL) instead of FedAvgGIN federated · NCI1 (30 clients)average accuracy: 0.6474 -> 0.7326Improved
use gradient-sequence DTW clustering (GCFL+) instead of last-gradient clustering (GCFL)GCFL · NCI1 (30 clients)average accuracy: 0.7326 -> 0.7422Improved
use GCFL+ instead of GCFLGCFL · PROTEINS (10 clients)average accuracy: 0.7739 -> 0.7776Improved
use GCFL+ instead of GCFLGCFL · IMDB-BINARY (10 clients)average accuracy: 0.8256 -> 0.8299Improved
use GCFL+ instead of GCFLGCFL · MOLECULESaverage accuracy: 0.7661 -> 0.7745Improved
use GCFL+ instead of GCFLGCFL · BIOCHEMaverage accuracy: 0.7172 -> 0.7312Improved
use GCFL+ instead of GCFLGCFL · MIXaverage accuracy: 0.7056 -> 0.7121Improved
apply FedAvg instead of local trainingGIN · MIX (13 clients)average accuracy: 0.7001 -> 0.6886Hurt
apply FedAvg instead of local trainingGIN · NCI1 (30 clients)average accuracy: 0.6468 -> 0.6474No clear effect
standardize the gradient-norm series in GCFL+ (ablation)GCFL+ · single and multi datasetperformance gainNo clear effect
vary clustering thresholds epsilon1 and epsilon2 within feasible ranges (ablation)GCFL/GCFL+ · MOLECULESaverage 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