HE, Xinlei - HKUST Class Schedule & Quota

Class Schedule & Quota
2025-26 Fall 2024-25 Summer 2024-25 Spring 2021-22 Winter
2025-26 Fall
Common Core Course
The class information on this page will normally be updated every 15 to 20 minutes during class enrollment period.
» Student Information System (SIS)» PG Program/Course Catalog
Notes & Links
This service was first adapted from a student contribution in 2011-12 Fall. The efforts of all concerned are duly acknowledged.

AIAA AMAT BSBE CMAA CNCC CNGF DLED DSAA EOAS FTEC FUNH INFH INTR IOTA IPEN LANG MICS MSSM PDEV PLED ROAS SEEN SMMG SOCH SYSH UCMP UCUG UFUG UGOD

COURSE INFO

VECTOR	[3-0-0:3]
DESCRIPTION	This course explores the intersection of privacy, security, and the rapidly evolving landscape of foundation models—large-scale AI systems like ChatGPT and DeepSeek that power modern applications. Students will gain foundational knowledge of these models, including their architecture, training methodologies, and real-world applications. The course will then dive into emerging threats, such as data privacy risks (e.g., model inversion, membership inference) and security vulnerabilities (e.g., adversarial attacks, poisoning). We will examine state-of-the-art defense strategies, including robust training techniques, cryptographic protections, and privacy-preserving frameworks, while addressing ethical, regulatory, and practical challenges. By the end of the course, students will be equipped to identify, analyze, and mitigate privacy and security risks in foundation models, balancing innovation with responsible deployment.

VECTOR

[3-0-0:3]

DESCRIPTION

This course explores the intersection of privacy, security, and the rapidly evolving landscape of foundation models—large-scale AI systems like ChatGPT and DeepSeek that power modern applications. Students will gain foundational knowledge of these models, including their architecture, training methodologies, and real-world applications. The course will then dive into emerging threats, such as data privacy risks (e.g., model inversion, membership inference) and security vulnerabilities (e.g., adversarial attacks, poisoning). We will examine state-of-the-art defense strategies, including robust training techniques, cryptographic protections, and privacy-preserving frameworks, while addressing ethical, regulatory, and practical challenges. By the end of the course, students will be equipped to identify, analyze, and mitigate privacy and security risks in foundation models, balancing innovation with responsible deployment.

IOTA 6910H - Privacy and Security Threat to Foundation Models (3 units)

Section	Date & Time	Room	Instructor	Quota	Enrol	Avail	Wait	Remarks
L01 (6172)	Th 06:30PM - 09:20PM	Rm 202, W4	HE, Xinlei	30	27	3	0