Instructor: Michael Dinitz
Lecture: T Th 9am - 10:15am, Hodson 311
Office Hours: By appointment
Teaching Assistant: Shruthi Prusty

Resources

• Syllabus
• Textbook: The Algorithmic Foundations of Differential Privacy, by Cynthia Dwork and Aaron Roth
• Online discussion board: Courselore. To join, use this invite link

---

Schedule

Date	Topic	Reference	Notes
Jan 21	Introduction, Definitions	Chapters 1, 2	Lecture Notes, Video
Jan 23	Definitions, Laplace Mechanism	Chapters 2, 3.1-3.3	Lecture Notes, Video
Jan 28	Laplace, Report-Noisy-Max	Chapter 3.3	Lecture Notes, Video
Jan 30	Exponential Mechanism	Chapter 3.4	Lecture Notes, Video
Feb 4	Group Privacy, Basic Adaptive Composition	Chapters 2.3, 3.5	Lecture Notes, Video, Northeastern Notes
Feb 6	Advanced Composition	Chapter 3.5	Lecture Notes, Video, Northeastern Notes, Toronto Notes
Feb 11	Sparse Vector Technique	Chapter 3.6	Lecture Notes, Video, HW1 out
Feb 13	Linear Queries and SmallDB	Chapter 4.1	Lecture Notes, Video
Feb 18	SmallDB, Private MWU	Chapter 4.1, 4.2	Lecture Notes, Video, AHK Survey on MWU
Feb 20	Private MWU	Chapter 4.2	Lecture Notes, Video
Feb 25	Nets and Iterative Construction	Chapter 5.1, 5.2	Lecture Notes, Video
Feb 27	Iterative Construction and the Median Mechanism	Chapter 5.2	Lecture Notes, Video
Mar 4	Binary Tree Mechanism	Lecture Notes from BU/NEU	Lecture Notes, Video
Mar 6	Local Sensitivity and Propose-Test-Release	Vadhan Section 7.3, Lecture Notes from Waterloo	Lecture Notes
Mar 11	Stable Histograms, Private Local Sensitivity, Smooth Senstivity	Vadhan Section 7.3	Lecture Notes, Video
Mar 13	Factorization and Projection	Lecture Notes from NEU/BU	Lecture Notes, Video
Mar 25	Other notions of DP: Renyi DP and zCDP	RDP paper, zCDP paper	Lecture Notes, Video
Mar 27	DP SGD	DP-SGD paper, BU/NEU Lecture Notes, Waterloo Lecture Notes, Google DP-fy ML	Lecture Notes, Video

---

Assignments

Please submit all assignments using Gradescope (entry code D3524N).

• Homework 1 (due February 20): PDF, LaTeX
• Homework 2 (due March 13): PDF, LaTeX
• Homework 3 (due April 3): PDF, LaTeX

Final Project Information

This class will have a final project, whose exact form is flexible and mostly up to you, subject to approval by the instructor. You can work in small groups if you want. Possible ideas include:

• Doing new research in algorithmic aspects of differential privacy.
• Doing research on a problem in your research area that can be combined with differential privacy (e.g., machine learning with privacy, HCI with privacy, etc.)
• Preparing a lecture / presentation / writeup on some aspect of differential privacy that we did not cover in class.

Other Resources

Other textbooks:

• The Complexity of Differential Privacy, by Salil Vadhan

Other classes:

• Algorithmic Foundations of Data Privacy, Penn
• Algorithms for Private Data Analysis, Waterloo