May 14, 2025
Apostol Vassilev
This presentation is a primer about the landscape of Adversarial Machine Learning (AML) including common types of AML attacks and appropriate mitigations. AML risk occurs because of "deliberate actions by motivated adversaries." This is one of two types of risks associated with using AI/ML models. The other is inherent risk (e.g., modeling errors, implementation flaws) and not covered in this presentation.