Meets 3:05-4:20pm, Tu/Th in Hudson Hall 212
Instructor: Dr. Emily Wenger Office: Fitzpatrick 2535 Office hours: 1 - 3pm, Tuesdays (Use Calendly to schedule) Email: [email protected]
TA: Sohini Saha Office: TBD Office hours: TBD Email: [email protected]
This graduate-level course will cover important developments in the dynamic field of AI security and privacy. It will provide students with a comprehensive understanding of the vulnerabilities inherent in AI models and novel security issues of generative AI models. We will start with well-understood vulnerabilities of AI models, such as adversarial attacks, backdoor attacks, and membership inference attacks, then move on to nascent security issues associated with widescale adoption of generative AI. By the end of the course, students will have a thorough understanding of known AI security and privacy issues.
Each class period, we will discuss papers on a topic related to AI security and privacy. After the first week of class, these discussions will be facilitated by students, who will present on the required readings and then lead a discussion, to which all students are expected to contribute. Students will also undertake a full-semester research project on a specific area of interest within AI security and privacy and will present their findings at the end of the semester. Students will be graded on class participation, the lectures/discussions they lead, engagement with readings/presentations during class sessions, and the quality of research they produce at the end of the semester. See “Grading” section below.
Paper presentation + leading discussions: Starting the second week of class, a team of 2 students will present the papers assigned for reading each week. Presentation assignments can be found here. During the paper presentation, the instructor will use the following rubric to grade presentations. Presenters will be graded individually on their ability to engage the class for discussion.
Weekly discussion questions: Each class session, each student is expected to submit one question for each of the 2 papers assigned for reading. These questions will be used to bootstrap the discussion after the presentation. Your questions should be submitted via Gradescope (see Canvas) by 2pm on the day of class, to allow presenters to incorporate these into their planned discussion. A rubric offering guidance on how to ask informative questions can be found here.
Research projects: Students will work on a course-long research project. Each project will be presented in class on November 26th (~10 minutes/presentation), and students will submit a 5 page writeup using the standard 2-column ACM template in Latex. In addition the presentation and writeup, note that a problem statement is due Sept 10th by beginning of class. Students are encouraged to consult the teaching team regularly to ensure progress is made throughout the semester. A rubric offering guidance on how the teaching team will evaluate research projects can be found here. Presentations will be graded using this rubric.
Important deadlines for research project:
Rubrics can be found here: https://docs.google.com/document/d/1YwbI4jxja1OuXO07YQSDq29w2UXSlXRRJ-OHCC1da5E/edit?usp=sharing
Class participation: 30% Class attendance: 20% Research Project: 25% Presentation/discussion leadership: 25%