Machine learning models are increasingly being deployed in critical applications, yet they remain vulnerable to subtle manipulations that can cause dramatic failures. In this course, Introduction to Adversarial AI, you'll learn to identify and understand the primary ways adversaries can attack modern AI systems. First, you'll explore the fundamental concepts behind adversarial examples, including perturbations, evasion attacks, and poisoning techniques. Next, you'll discover how to use industry-standard tools like CleverHans and ART to simulate real attacks on neural networks. Finally, you'll learn how black-box models can be reverse-engineered through model extraction techniques. When you're finished with this course, you'll have the skills and knowledge of adversarial AI needed to better understand the security vulnerabilities in your machine learning systems and take the first steps toward protecting them.
Overview
Machine learning models are increasingly being deployed in critical applications, yet they remain vulnerable to subtle manipulations that can cause dramatic failures. In this course, Introduction to Adversarial AI, you'll learn to identify and understand the primary ways adversaries can attack modern AI systems. First, you'll explore the fundamental concepts behind adversarial examples, including perturbations, evasion attacks, and poisoning techniques. Next, you'll discover how to use industry-standard tools like CleverHans and ART to simulate real attacks on neural networks. Finally, you'll learn how black-box models can be reverse-engineered through model extraction techniques. When you're finished with this course, you'll have the skills and knowledge of adversarial AI needed to better understand the security vulnerabilities in your machine learning systems and take the first steps toward protecting them.
Taught by
Goran Trajkovski
