Current students

Adversarial Machine Learning (AML) is an effective framework to study the robustness of Machine Learning (ML) models in adversarial settings. Indeed, ML techniques may fail in domains such as cybersecurity, where the data is poisoned by an opponent. This project aims to formulate the AML framework using Game Theory to handle the interplay between the learner and the attacker. The effectiveness of the proposed framework will be assessed by designing novel methods for cybersecurity applications.