Calendar & Syllabus

Week 1 - Introduction and Privacy Attacks

Jul 29
Introduction & Logistics
Slides   Reading - Narayanan & Shmatikov (2008)
HW 0 released HW0 PDF
Jul 31
Privacy Attacks - Membership Inference
Required - Shokri et al. (2017), Carlini et al. (2021) Extra - Mireshgallah et al. (2022)
Aug 1
Privacy Attacks - Data Reconstruction
Required - Carlini et al. (2021) Extra - Nasr et al. (2023), Haim et al. (2022)

Week 2 - Differential Privacy Basics I

Aug 5
Lab PyTorch Review
Additional tutorials
Aug 7
Introduction to differential privacy (DP)
Reading - Sec. 1.4-1.6 of Vadhan
Aug 8
DP - basic composition
Reading - Sec. 2 of Steinke
Aug 9
HW 0 due

Week 3 - Differential Privacy Basics II

Aug 12
Concentration of Measure - part 1 (No lab)
Duchi’s notes
Aug 14
Concentration of Measure - part 2
Sec. 1 - 3 of Rivasplata’s note
Aug 15
No class (Independence Day)
HW 1 released

Week 4 - Differential Privacy Basics III

Aug 19
Lab Privacy Accounting
dp-accounting library
Aug 21
Concentrated DP & advanced composition
Reading - Sec. 4.1-4.2 of Steinke
Aug 23
Amplification by Subsampling
Reading - Sec. 6.1-6.3 of Steinke

Week 5 - Learning with DP I

Aug 26
Lab Per-sample gradients in PyTorch
Tutorial
Aug 28
Rényi DP - Subsampling + Composition
Reading - Sec. 6.4 of Steinke and Mironov et al. Optional: Sec. 6.5 and 6.4 of Steinke
Aug 29
Stochastic gradient descent with DP + Practical considerations
Reading - Sec. 4.2 of Ponomareva et al. (Sec. 5 is also strongly recommended)
Aug 30
HW 1 due
HW 2 released

Week 6 - Learning with DP II

Sept 2
Lab DP-SGD
Sept 4
Class Cancelled, make-up to be announced later Theoretical Analysis of DP-SGD
Sept 5
Private learning with Correlated Noise - part 1
Reading - see Piazza

Week 7 - Learning with DP III

Sept 9
Lab Correlated noise mechanisms
Sept 11
Private learning with Correlated Noise - part 2
Reading - see Piazza
Sept 12
Other DP-learning methods: perturbation, ensembling
Objective perturbation: Chaudhuri et al., PATE: Papernot et al. and Papernot et al.
Sept 15
HW 2 due

Week 8 - Projects & Review

Sept 16
Holiday   Project Discussions
Call for Projects Released
Sept 18
Project Discussions   DP’s protections for reconstruction
Sept 19
Homework Review   Reconstruction Protection via Fisher Information

Week 9 - No class

Sept 23
No class: work on project proposals
Sept 25
No class: work on project proposals
Sept 26
No class: work on project proposals
Sept 27
Project Proposals due

Week 10 - Review & Midterm

Sept 30
Midterm Review
Oct 2
Holiday
Oct 3
Midterm
Oct 4
HW3 released

Week 11 - Federated Learning and Privacy with Distributed Data

Oct 7
Lab Auditing DP
Oct 9
Federated Learning & Privacy
Required - Interactive blog post
Oct 10
Local & Distributed DP
Required - Kairouz et al.
Oct 31
HW 3 due

Week 12 - Protecting Against Data Reconstruction

Oct 14
(Online) Guest Lecture by Ashwinee Panda on Private In-Context Learning in LLMs
Paper
Oct 16
DP’s protections for reconstruction
Stock et al.
Oct 17
Reconstruction Protection via Fisher Information
Guo et al.

Week 13 - Privacy in Generative AI

Oct 21
Project Office Hours
Oct 23
Copyright Protection in Generative AI
Vyas et al.
Oct 24
Synthetic text generation (+ exponential mechanism)
Amin et al. (Smith’s notes for the exponential mechanim)
Oct 27
Project midpoint report due

Week 14 - Advanced Topics

Oct 28
(Online) Guest lecture by Eugene Bagdasaryan on LLM Privacy and Contextual Integrity
Paper
Oct 30
Unlearning
Chien et al.
Oct 31
Holiday

Week 15 - Project Presentations

Week 16 - Buffer / End-sem Week

Nov 15
Project final report due at noon