Project: Fairness Analysis in Anomaly Detection
Description
In anomaly detection, we aim to identify unusual instances in different applications, including malicious users detection in OSNs, fraud detection, and suspicious bank transaction detection. Most of the proposed anomaly detection methods are dependent on network structure as some specific structural pattern can convey abnormal behavior. Unfairness in such systems might affect some particular communities, for example, targeting users from a specific community while identifying suspicious users. Davidson and Ravi [2] compared five classic anomaly detection methods and showed that their outputs are unfair; however, these works might mislead someone to conclude that their results are fair, especially when the number of outliers and the number of protected status variables are small. The such analysis raises the question of whether anomalous nodes and links detection methods are fair or not for different protected groups in complex networks. If not, then anomaly detection methods for complex network data should address these issues and focus on the fairness of all protected groups.
Saxena, Akrati, George Fletcher, and Mykola Pechenizkiy. "FairSNA: Algorithmic Fairness in Social Network Analysis." arXiv preprint arXiv:2209.01678 (2022).
Ian Davidson and Selvan Suntiha Ravi. 2020. A framework for determining the fairness of outlier detection. In ECAI 2020. IOS Press, 2465–2472.
Bin Zhou, Jian Pei, and WoShun Luk. 2008. A brief survey on anonymization techniques for privacy preserving publishing of social network data. ACM Sigkdd Explorations Newsletter 10, 2 (2008), 12–22.
Details
- Supervisor
-
Mykola Pechenizkiy
- Secondary supervisor
-
Akrati Saxena
- Interested?
- Get in contact