Privacy in Statistics and Machine Learning
Lecture videos (from 2021) can be found in this Google Drive folder.Spring 2025
| Date | Topics | Deliverables |
|---|---|---|
| Tue 1/21 |
Lecture 1: Course Overview Video: Watch these two MinutePhysics videos (1, 2). They introduce some of the course topics in the context of the US Census. Lecture Slides: pdf In-class Exercises: pdf (We did not get to these in class. They are posted for reference.) |
|
|
Thu 1/23 |
Lecture 2:
Reconstruction Attacks
I
Video: part 1, part 2 Lecture Notes: pdf that goes with videos (up through 2.3). Additional optional reading on reconstruction:
In-class Exercises: pdf |
|
| Tue 1/28 | Lecture 3:
Reconstruction Attacks
II Lecture Notes: pdf (2.4 onward) Video: part 1, part 2 In-class Exercises: pdf |
|
| Thu 1/30 |
Lecture 4: Differential Privacy Fundamentals I Lecture Notes: pdf Lecture Slides (in person lecture): pdf Video (from 2021, for reference): part 1, part 2, part 3, part 4 |
Homework 1 out (pdf), due Sunday February 16, 2025. |
| Tue 2/4 |
Lecture 5: Differential Privacy Fundamentals II Lecture Notes: pdf Video: part 1, part 2, part 3 In-class Exercises: pdf |
|
| Thu 2/6 |
Lecture 6: Exponential Mechanism and Report Noisy Max Lecture Notes: pdf Video: part 1, part 2, part 3, part 4, In-class Exercises: pdf |
|
| Tue 2/11 |
Lecture 7:
The Binary Tree Mechanism Lecture Notes: pdf Video: part 1, part 2 In-class Exercises: pdf |
|
| Thu 2/13 |
Lecture 8: Approximate DP In-person lecture; no advanced reading required. Lecture Notes: pdf Video: part 1, part 2 In-class Exercises: N/A |
|
| Tue 2/18 | No class (Monday schedule) | |
| Thu 2/20 |
Lecture 9: Advanced Composition Lecture Notes: pdf Video: part 1, part 2 In-class Exercises: pdf |
|
| Tue 2/25 |
Lecture
10: Recap
(No lecture notes or videos.) In-class Exercises: pdf |
|
| Thu 2/27 |
Lecture
11: Private Empirical Risk Minimization Lecture Notes: pdf Video: part 1, part 2 In-class Exercises: pdf |
|
| Tue 3/4 |
Lecture
12: Private Gradient Descent Lecture Notes: pdf Video: part 1, part 2 In-class Exercises: pdf |
|
| Thu 3/6 |
Lecture
13: Factorization Mechanisms Lecture Notes: pdf (up to 2.3) Video: part 1, part 2 In-class Exercises: pdf |
Homework 2 out (pdf), due Friday, March 21, 2025. |
| Tue 3/12 | No class (Spring break) | Project info out (pdf). Proposals due Wednesday, March 26, 2025 |
| Thu 3/14 | No class (Spring break) | |
| Tue 3/18 |
Lecture 14: DP-FTRL
Slide notes: pdf Based on Kairouz, McMahan, Song, Thakkar, Thakurta, Xu (NeurIPS 2021) (as well as a few subsequent papers). |
|
| Thu 3/20 |
Lecture
15: The Projection Mechanism Lecture Notes: pdf (2.3 onward) Video: part 1 In-class Exercises: pdf |
|
| Tue 3/25 |
Lecture 16: Online
Learning and Multiplicative Weights Lecture Notes: pdf (Sections 1 to 3.1) Video: part 1, part 2, part 3, part 4 In-class Exercises: pdf |
|
| Thu 3/27 |
Lecture 17: Synthetic Data
Generation and Online Learning Lecture Notes (hand-written): pdf We covered this material in a live lecture, but here are vidoes from 2021: part 1, part 2, part 3, part 4, part 5, part 6. |
|
| Tue 4/1 |
Lecture 18: The Limits of Privacy Lower Bounds from Reconstruction and Membership Inference Attacks (in person) Lecture Notes: pdf |
|
| Thu 4/3 |
Lecture 19: Two-Player Zero-Sum Games I Lecture Notes: pdf (up to the end of 1.2) Video: part 1 (Only one video for Lec 19.) In-class excercises: pdf |
|
| Tue 4/8 |
Lecture 20: Two-Player Zero-Sum Games II Lecture Notes: pdf (1.3 onward) Video: part 1 part 2 part 3 In-class excercises: pdf |
|
| Thu 4/13 |
Lecture 21: Private Statistical Inference Lecture slides: pdf |
|
| Tue 4/15 |
Lecture 22: A Simple
Analysis of Robust Membership
Inference Attacks
Hand-written Lecture Notes: pdf |
|
| Thu 4/17 |
Lecture 23: More on Membership Inference and Memorization
Plots on membership inference: Colab Lecture Slides: pdf |
|