Syllabus Overview
Course Description
This course will focus on the theory of differential privacy, the gold-standard framework for reasoning about privacy in data analysis. We will start with foundational concepts and definitions, focusing on the question of what it means to conduct private data analysis. We will acquaint ourselves with the basic toolkit of private algorithm design and build up to state-of-the-art approaches in machine learning. Particular attention will be paid to the theory of private statistical inference.
The bulk of the course will be theoretical and proof-based. It is designed for graduate students and strong undergraduate students with a background in probability, statistics, and algorithms. Grading will be based on mathematical exercises, lightweight coding assignments, in-class participation, lecture scribing, and a final project.
Course Materials & Resources
In this class, we will commonly refer to the following open-access texts. You can find additional resources, including high-quality blog posts and pointers to classes at many institutions, at differentialprivacy.org.Assessments & Grading
- 50% Homework (4-5 assignments)
- 30% Final Project
- 10% Class Attendance and Participation
- 10% Lecture Scribing
Course Policies
- Attendance: Lecture attendance is mandatory and participation is expected.
- Generative AI: Students are allowed to use AI tools in their assignments and scribing. As always, students bear responsibility for the content of their submissions. If you submit an AI-generated solution with a reckless disregard for correctness, the instructor will take it as a personal affront.