CS-573-1: Data Privacy and Security - Fall 2024

Quick Links:  Course Schedule | Project Requirements   

Lecture: MW 11:30-12:45pm MSC W303

Instructor: Li Xiong (lxiong@emory.edu)
Office Hours: M 2:30-3:30pm W 12:45-1:45pm or by appointment, MSC E412

Co-Instructor:  Ruixuan Liu (ruixuan.liu2@emory.edu)
Office Hours: M 3:00-4:00pm F 12:00-13:00pm or by appointment, MSC N412

TA:  Toan Tran (viet.toan.tran@emory.edu)
Office Hours: Wed. 9:30 - 11:30 am (N414 MSC) or by appointment

Overview

This course will introduce students to data privacy and security issues and techniques in the context of AI and machine learning (ML). The main topics include privacy attacks on ML models, privacy enhancing technology for building ML models such as statistical privacy (differential privacy), federated learning, data synthesization, machine unlearning, and adversarial attacks on ML models such as adversarial example attacks, data poisoning attacks, backdoor attacks and corresponding defense techniques for building robust ML models. We will also study the privacy and security risks and defenses of the emerging large language models (LLMs).

Readings

There are no required textbooks. The class is based on seminal and recent research papers and selected book chapters. Here is a continuously maintained recommended reading list.

Prerequisites

Familiarity with Python and prior courses in algorithms and machine learning are required.

Assignments

There will be 3 programming assignments. You will be implementing existing privacy attacks, privacy-enhancing algorithms, adversarial attacks and/or robust algorithms.

Late policy

You have 6 flexible "24-hour late days" which you can use for any programming assignments. You can use up to 3 late days per single assignment.  They will be automatically deducted when the submission is late.  

Exam

There will be one in-class open-notes midterm exam and no final exam.

Project

The course includes a substantial course project. Students work in teams of 2-3 people.  Project deliverables include project proposal, in-class literature review presentation, in-class project presentation, and final project report/deliverable.  More details here. 

Grading