CurvGAD adds Ollivier-Ricci curvature reconstruction and Ricci-flow regularization to a graph autoencoder, improving node-level anomaly detection AUROC by up to 6.5% over state-of-the-art baselines on 10 datasets.
Constant curvature graph convolutional networks
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
CurvGAD: Leveraging Curvature for Enhanced Graph Anomaly Detection
CurvGAD adds Ollivier-Ricci curvature reconstruction and Ricci-flow regularization to a graph autoencoder, improving node-level anomaly detection AUROC by up to 6.5% over state-of-the-art baselines on 10 datasets.