AI Paper Atlas

Research map / Graph learning / Graph self-supervised learning

MV-FGAD: Towards Efficient and Effective Federated Graph Anomaly Detection via Multi-view Learning

Junyi Yan, KE LIANG, Hao Yu, Meng Liu, Hao Tan, Tianrui Liu, Jun-Jie Huang, Xinwang Liu

ICML 2026 · Graph learning › Graph self-supervised learning · approach: Graph anomaly detection

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
use Mahalanobis distance instead of Euclidean distance for scoring (ablation)MV-FGAD · AmazonAUROC: 71.11 -> 78.33Improved
use Mahalanobis distance instead of cosine similarity for scoring (ablation)MV-FGAD · RedditAUROC: 55.66 -> 59.27Improved
use Mahalanobis distance instead of cosine similarity (ablation)MV-FGAD · TolokersAUPRC: 23.31 -> 23.56Improved
combine any two views of AAL, SAL, WADL (ablation)MV-FGAD · Reddit, Tolokers, AmazonAUROCImproved
add all three views (AAL, SAL, WADL) (ablation)MV-FGAD · AmazonAUROC: 75.62 -> 78.33Improved
add weak anomaly-aware dynamic learning view (ablation)MV-FGAD · multiple datasetsAUROCImproved
use mean aggregation of multi-view scores (ablation)MV-FGAD · multiple datasetsstabilityImproved
use MV-FGAD instead of federated GAD baselinesMV-FGAD · Amazon-allAUPRC: 20.61 -> 40.53Improved
use MV-FGAD instead of LG-FGADMV-FGAD · YelpChiAUPRC: 33.36 -> 22.60Mixed
set alpha and beta to 0.5 (ablation)MV-FGAD · multiple datasetsAUROCImproved

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