ProtoGLAD detects graph-level anomalies by contrasting with nearest normal prototype graphs discovered via point-set kernels. It iteratively clusters normal graphs for unsupervised detection. Provides human-interpretable explanations outperforming black-box methods.
Key Points
- 1.Prototype-based contrast explanations
- 2.Iterative normal cluster discovery
- 3.Competitive on real datasets
Impact Analysis
Boosts reliability for real-world GLAD deployment with concrete graph references.
Technical Details
Point-set kernel for prototypes; anomalies as distant from all clusters.