COMS 5580X: Foundations of Generative AI for Cybersecurity
(Dual-listed with COMS 4580X).
Credits: 3. Contact Hours: Lecture 3.
Repeatable, maximum of 3 credits.
Prereq: Graduate Standing or Permission of Instructor
Foundations of generative AI with emphasis on cybersecurity principles, threats, and defenses. Core model families such as VAEs, GANs, diffusion, flows, energy-based models, and LLMs and how they are trained and deployed in adversarial settings. Key topics include representation learning, transformers, prompting and agentic AI, offensive uses (e.g., exploit generation, prompt injection, poisoning, backdoors, exfiltration, privacy leakage), and defensive methods such as interpretability, drift/robustness evaluation, provenance/authentication, and secure hardening.
(Typically Offered: Fall)