Adversarial Machine Learning

Adversarial Machine Learning, 1st Edition functions as a textbook dedicated to analyzing security vulnerabilities in artificial intelligence systems. The text establishes an operational foundation by examining security risks across the machine learning pipeline, covering training, deployment, and inference stages.

The content categorizes major attack methodologies into distinct thematic areas. Coverage includes evasion attacks that manipulate input data, backdoor and trojan attacks that insert hidden behaviors, and generative AI attacks such as prompt injection. To counter these threats, the text details technical defensive concepts including certified defenses and risk-aware system design.

This textbook provides targeted technical material for developers, security engineers, and AI decision-makers. The content prepares these professionals to analyze model vulnerabilities and evaluate defensive strategies across enterprise AI workflows.