Anomaly Detection on Networks is a Question of Context and Scale
L. Gutierrez-Gomez, A. Bovet, and J.C. Delvenne
Ercim News, no. 122, pp. 49-50, 2020
Anomaly detection is an important problem in data mining with diverse applications in multiple domains. Anomalies, also known as outliers, can be defined as individual objects with patterns or behaviours that differ starkly from a background property. Examples of applications include fraud detection in finance, detection of faults in manufacturing, identifying fake news in social media, or web spam detection. Anomalies in real problems may lead to enormous economic, social, or political costs and are often difficult to find, mainly because they are scarce and unknown a priori. Therefore, efficient detection of anomalies may bring significant value to people, companies, and authorities.