Learning Discriminative and Generalizable Anomaly Detector for Dynamic Graph with Limited Supervision

📰 ArXiv cs.AI

Learn to build a discriminative and generalizable anomaly detector for dynamic graphs with limited supervision, improving detection accuracy in real-world applications

advanced Published 2 Jun 2026

Action Steps

Build a graph neural network (GNN) to learn node and edge representations
Apply self-supervised learning techniques to pre-train the GNN on unlabeled data
Fine-tune the pre-trained GNN on limited labeled anomaly data
Evaluate the detector's performance using metrics such as precision, recall, and F1-score
Compare the results with existing unsupervised and semi-supervised methods

Who Needs to Know This

Data scientists and ML engineers working on graph-based anomaly detection tasks can benefit from this research, as it provides a novel approach to handling limited labeled data

Key Insight

💡 Self-supervised learning can help pre-train a graph neural network to improve anomaly detection accuracy in dynamic graphs

Key Takeaways

Learn to build a discriminative and generalizable anomaly detector for dynamic graphs with limited supervision, improving detection accuracy in real-world applications

Full Article

Title: Learning Discriminative and Generalizable Anomaly Detector for Dynamic Graph with Limited Supervision

Abstract:
arXiv:2602.20019v2 Announce Type: replace-cross Abstract: Dynamic graph anomaly detection is critical for many real-world applications but remains challenging due to the scarcity of labeled anomalies. Existing methods are either unsupervised or semi-supervised: unsupervised methods avoid the need for labeled anomalies but often produce ambiguous boundary, whereas semi-supervised methods can overfit to the limited labeled anomalies and generalize poorly to unseen anomalies. To address this gap, w

Read full paper → ← Back to Reads