KI-Glossar
Das vollständige Wörterbuch der Künstlichen Intelligenz
Graph Anomaly Detection
Set of techniques aimed at identifying abnormal patterns, nodes or relationships in graph data structures compared to expected or normal behavior.
Node Anomaly Detection
Process of identifying nodes in a graph that exhibit structural or attribute characteristics significantly different from the majority of other nodes.
Edge Anomaly Detection
Technique for detecting unusual edges or connections in a graph, often based on abnormal weights, frequencies or connection patterns.
Subgraph Anomaly Detection
Method for identifying complete subgraph structures exhibiting statistical or structural properties divergent from the global graph.
Graph Neural Networks for Anomaly Detection
Deep learning architecture adapted to graph structures using message passing mechanisms to detect anomalies through representation learning.
Spectral Anomaly Detection
Approach based on the analysis of eigenvalues and eigenvectors of the adjacency or Laplacian matrix to identify structural anomalies in graphs.
Graph Embedding for Anomaly Detection
Transformation of graph entities into low-dimensional vector spaces enabling the application of classical anomaly detection algorithms.
Temporal Graph Anomalies
Detection of abnormal behaviors in dynamic graphs where relationships and attributes evolve temporally, requiring spatio-temporal analysis.
Attributed Graph Anomaly Detection
Detection of anomalies in graphs enriched with attributes on nodes and edges, combining structural and semantic information for robust detection.
Graph Community Anomaly Detection
Identification of abnormal communities or clusters with unusual internal or external connection densities compared to typical communities.
Graph Outlier Detection
Systematic process of identifying extreme or deviating observations in graph data that may indicate errors, fraud, or malicious behavior.
Graph Pattern Anomaly Detection
Search for abnormal recurring patterns or schemes in graphs, often based on appearance frequencies or unusual topological structures.
Graph Centrality Anomaly Detection
Anomaly detection based on the analysis of centrality measures (betweenness, closeness, eigenvector) to identify nodes with abnormally high or low influence.
Graph Degree Anomaly Detection
Identification of nodes with abnormally high or low connection degrees compared to the expected distribution in the graph.
Graph Neighborhood Anomaly Detection
Technique analyzing local characteristics of node neighborhoods to detect anomalies based on unusual neighborhood structures.
Dynamic Graph Anomaly Detection
Detection of anomalies in evolving graphs where nodes and edges appear, disappear, or change over time, requiring adaptive algorithms.
Graph Structure Anomaly Detection
Purely topological analysis of the graph to identify structural anomalies independent of attributes, based on connectivity and morphological properties.
Graph Feature Anomaly Detection
Anomaly detection based on the extraction and analysis of specific graph features such as triangles, paths, or unusual local patterns.