A New Type of Anomaly Detection Problem in Dynamic Graphs: An Ant Colony Optimization Approach

• Published in: Bioinspired Optimization Methods and Their Applications Vol. 13627; pp. 46 - 53
• Main Authors: Tasnádi, Zoltán, Gaskó, Noémi
• Format: Book Chapter
• Language: English
• Published: Switzerland Springer International Publishing AG 2022; Springer International Publishing
• Series: Lecture Notes in Computer Science
• Subjects: ACO; Anomaly detection; Community detection; Heaviest; subgraph
• ISBN: 303121093X; 9783031210938
• ISSN: 0302-9743; 1611-3349
• DOI: 10.1007/978-3-031-21094-5_4

Anomaly detection has gained great attention in complex network analysis. Every unusual behavior in a complex system can be viewed as an anomaly. In this article, we propose a new anomaly type in dynamic graphs, an existing community-based anomaly detection problem combined with the heaviest k-subgraph problem. Searching the heaviest subgraphs in dynamic graphs viewed as an anomaly problem can give new insights into the studied dynamic networks. An ant colony optimization algorithm is proposed for the heaviest k-subgraph problem and used for the community detection problem. Numerical experiments on real-world dynamic networks are conducted, and the results show the importance of the proposed problem and the potential of the solution method.