G2A2: Graph Generator with Attributes and Anomalies
Saikat Dey, Sonal Jha, Wu-chun Feng · 2024
Many data-mining applications use dynamic attributed graphs to represent relational information; but due to security and privacy concerns, there is a dearth of publicly available datasets that can be represented as dynamic attributed graphs. Even when such datasets are available, they do not have ground truth that can be useful for classification problems, e.g., anomaly detection. Thus, researchers commonly generate synthetic graphs using either statistical or deep generative (DG) methods. However, neither approach produces ground truth. Statistical methods struggle to replicate intricate patterns found in real-world dynamic attributed graphs, while DG methods require a significant number of graphs for training.