Abstract
AbstractLearning from data streams is a hot topic in machine learning and data mining. This article presents our recent work on the topic of learning from data streams. We focus on emerging topics, including fraud detection and hyper-parameter tuning for streaming data. The first study is a case study on interconnected by-pass fraud. This is a real-world problem from high-speed telecommunications data that clearly illustrates the need for online data stream processing. In the second study, we present an optimization algorithm for online hyper-parameter tuning from nonstationary data streams.
Publisher
Springer International Publishing
Reference5 articles.
1. Ali, M. A., Azad, M. A., Centeno, M. P., Hao, F., van Moorsel, A.: Consumer-facing technology fraud: Economics, attack methods and potential solutions. Future Generation Computer Systems, 100, 408–427 (2019)
2. Demšar, J.: Statistical comparisons of classifiers over multiple data sets. Journal of Machine Learning Research, 7(Jan), 1–30 (2006)
3. Laleh, N., Azgomi, M. A.: A taxonomy of frauds and fraud detection techniques. In International Conference on Information Systems, Technology and Management, pp. 256–267. Springer (2009)
4. Veloso, B., Gama, Malheiro, J. B., Vinagre, J.: Hyperparameter self-tuning for data streams. Information Fusion, 76, 75–86 (2021)
5. Veloso, B., Tabassum, S., Martins, C., Espanha, R., Azevedo, R., Gama, J.: Interconnect bypass fraud detection: a case study. Annals des Télécommunications, 75(9), 583–596 (2020)