A graph autoencoder with meta-path-level contrastive learning is proposed for unsupervised heterogeneous graph anomaly detection, reporting top AUC on DBLP, Aminer, and Yelp.
Deep graph learning for anomalous citation detection,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
EAGLE: Contrastive Learning for Efficient Graph Anomaly Detection
A graph autoencoder with meta-path-level contrastive learning is proposed for unsupervised heterogeneous graph anomaly detection, reporting top AUC on DBLP, Aminer, and Yelp.