Anomaly Detection in Dynamic Graphs via Transformer

Abstract

Detecting anomalies for dynamic graphs has drawn increasing attention due to their wide applications in social networks,e-commerce, and cybersecurity. Recent deep learning-based approaches have shown promising results over shallow methods.However, they fail to address two core challenges of anomaly detection in dynamic graphs: the lack of informative encoding forunattributed nodes and the difficulty of learning discriminate knowledge from coupled spatial-temporal dynamic graphs. To overcomethese challenges, in this paper, we present a novelTransformer-basedAnomalyDetection framework forDYnamic graphs (TADDY).Our framework constructs a comprehensive node encoding strategy to better represent each node’s structural and temporal roles in anevolving graphs stream. Meanwhile, TADDY captures informative representation from dynamic graphs with coupled spatial-temporalpatterns via a dynamic graph transformer model. The extensive experimental results demonstrate that our proposed TADDY frameworkoutperforms the state-of-the-art methods by a large margin on six real-world datasets.