Date |
Topic |
Reading/Reference |
Homeworks |
Jan 29 |
Introduction, Course Overview, Definition of Differential Privacy
|
Dwork-Roth, Ch. 1, Ch. 2 upto Def 4 | |
Jan 31 |
Randomized Response, Laplace Mechanism |
Dwork-Roth, Sec. 3.2-3.3.0 |
|
Feb 5 |
Understanding the Definition of DP |
Dwork-Roth, Ch. 2 and definition-notes.pdf | |
Feb 7 |
Composition Theorems |
Dwork-Roth, Sec 3.4 | HW 1 |
Feb 12 |
Exponential Mechanism and Answering Many Queries via Synthetic Data |
Dwork-Roth, Sec 3.3.1, Ch. 4 Intro, Sec. 4.1 | |
Feb 14 |
The Sparse Vector Technique |
Dwork-Roth, Sec. 3.5 | |
Feb 19 |
Answering Many On-line Queries: Private Multiplicative Weights |
Dwork-Roth, Sec. 4.2 | |
Feb 21 |
Attacks & Lower Bounds |
Dwork-Roth, Ch. 8 (except Thm 113) and packing.pdf | HW 1 due (Fri 2/22) HW 2 |
Feb 26 |
Alternatives to Worst-Case Sensitivity |
Dwork-Roth, Ch. 7 | |
Feb 28 |
Hardness of Generating Private Synthetic Data |
Ullman, Vadhan. "PCPs and the Hardness of Generating Private Synthetic Data." Sections 1,2, and 4.1 of syntheticdata.pdf | |
Mar 5 |
Class cancelled |
||
Mar 7 |
Class cancelled |
||
Mar 12 |
Finish Hardness of Synthetic Data |
Ullman, Vadhan. "PCPs and the Hardness of Generating Private Synthetic Data." Sections 3 and 4.2 of syntheticdata.pdf | Project topic ideas due (Wed 3/13) See project guidelines and project_ideas.pdf for more information |
Mar 14 |
Faster Algorithms for Marginal Queries |
Thaler, Ullman, Vadhan. "Faster Algorithms for Privately Releasing Marginals." (Can skip Sections 4.2, 4.3) fastermarginals.pdf | HW 2 due (Fri 3/15) HW 3 |
Mar 19 |
Spring Break |
||
Mar 21 |
Spring Break |
||
Mar 26 |
Graph Analysis: edge-level privacy |
Karwa et al. "Private Analysis of Graph Structure" | |
Mar 28 |
Graph analysis: node-level privacy |
Blocki et al. "Differentially Private Data Analysis of Social Networks via Restricted Sensitivity", and Sec 3-4.2 of Kasiviswanathan et al. "Analyzing Graphs with Node Differential Privacy" | |
Apr 2 |
Differential Privacy in Computational Learning Theory |
Kasiviswanathan et al. "What Can We Learn Privately?" Sections 1-3.1 | |
Apr 4 |
Differential Privacy in Computational Learning Theory, cont'd |
Kasiviswanathan et al. "What Can We Learn Privately?" Sections 4-5.1 | HW 3 due (Fri 4/5) |
Apr 9 |
Differential Privacy & Mechanism Design: paying for private data |
Ghosh & Roth "Selling Privacy at Auction" (especially Section 2) | |
Apr 11 |
Differential Privacy & Mechanism Design: privacy for standard mechanism design problems |
Chen et al. "Truthful Mechanisms for Agents that Value Privacy" Sections 1-5 | Project Proposals due (Fri 4/12) |
Apr 16 |
Alternate Definitions: crowd-blending privacy |
Gehrke et al. "Crowd-Blending Privacy" | |
Apr 18 |
Alternate Definitions: the pufferfish framework |
Kifer and Machanavajjhala "A Rigorous and Customizable Framework for Privacy" | |
Apr 23 |
Traitor-Tracing and the Complexity of Differental Privacy |
Ullman "Answering n2+o(1) Counting Queries with Differential Privacy is Hard" Sections 1-4 (If you're interested in how traitor-tracing schemes are constructed, feel free to read some or all of Section 5!) |
|
Apr 25 |
Conclusions |
None | |
Apr 30 |
Differential Privacy in Computational Learning Theory, cont'd |
Beimel et al. "Characterizing the Sample Complexity of Private Learning," Sections 1-3, 5 | |
May 2 |
No class |
Project Papers due (Fri 5/3) | |
May 7 |
Project Presentations 9:30-11:30 |
||
May 9 |
Project Presentations 9:30-11:30 |