Paper accepted at FPS 2022

Solayman’s first paper has been accepted for presentation at the 15th International Symposium on Foundations and Practice of Security (FPS 2022). The paper discusses the proposal of an evaluation framework for ML-based intrusion detection systems. It starts with a survey of recent works in the field of ML-based IDS and provides a comparison of their evaluation methodologies. We notice the lack of rigor with respect to ML best practices when public datasets are used. The corpus of works we studied often fail to anticipate common shortcomings such as unbalance, mislabelling or representativeness of the data. Additionally, although some novel evaluation methodologies were proposed in the field of intrusion detection, the community has been reluctant to apply them broadly. Our proposed framework attempts at combining the advances in both fields to improve the evaluation capabilities of intrusion detection systems that are based on learning models, often sensisitive to the quality of the data. If you happen to be in Ottawa at that time, come and let’s discuss!