SpecAE detects global and community anomalies in attributed networks by embedding nodes with a graph convolutional autoencoder and Laplacian sharpening, then scoring them with a Gaussian mixture energy.
Radar: Residual analysis for anomaly detection in attributed networks
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2019 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
SpecAE: Spectral AutoEncoder for Anomaly Detection in Attributed Networks
SpecAE detects global and community anomalies in attributed networks by embedding nodes with a graph convolutional autoencoder and Laplacian sharpening, then scoring them with a Gaussian mixture energy.