<ul><li>Graph anomaly detection is a crucial task that has been studied for decades, with recent deep learning-based methods showing promising results.</li><li>Efficiency is lacking in existing graph anomaly detection methods, especially for embedded devices.</li><li>A new model called EAGLE (Efficient Anomaly detection model on heterogeneous Graphs via contrastive LEarning) has been proposed to address this efficiency issue.</li><li>EAGLE utilizes contrastive learning to distinguish abnormal nodes from normal ones and outperforms existing methods on three different heterogeneous network datasets.</li></ul>

EAGLE: Contrastive Learning for Efficient Graph Anomaly Detection

Discover more