Lessons Learned from Evaluating the Robustness of Defenses to Adversarial Examples

This course covers the learning outcomes and goals of evaluating the robustness of defenses to adversarial examples in the context of machine learning classifiers. The individual skills taught include understanding adversarial examples, different defense mechanisms like adversarial training and input transformation, and how to evaluate the effectiveness of defenses. The teaching method involves a survey of past defense breaks, recommendations for thorough defense evaluations, and a comparison of evaluation methods in different security fields. The intended audience for this course includes researchers, practitioners, and students interested in machine learning security and adversarial attacks.