PhD Course - Differential Privacy Seminar

Title:
Differential Privacy Seminar 2018

Organisers:
Rasmus Pagh, Nina Mesing Stausholm Nielsen

Course website:
The following Google Docs document will be used to organize the course:
https://docs.google.com/document/d/1Bq5SZ37QjR04oRyWyK1ebWVoiAFBQXkJBbPsMa7AdZw/edi
t

Lecturers:
In each lecture, a participant will present a chapter/part of a chapter from the book The Algorithmic
Foundations of Differential Privacy by Cynthia Dwork and Aaron Roth.

Each participant must give at least one presentation during the seminar.

The lecturer for each lecture will be added to the Google Docs continuously.

Dates:
Lectures: October 22 nd to January 14 th
Project: January 14 th to March 1 st

Monday morning at 8:30.

Room:
TBA

Course description:
The seminar is organized such that over 10 lectures the participants will present topics from the
book The Algorithmic Foundations of Differential Privacy by Cynthia Dwork and Aaron Roth.
The lecturer should prepare a 30-45 minute presentation, find relevant exercises to discuss and
suggest related papers and/or project topics. The session will begin with the lecture, after which
there will be time for discussing the topic in general, exercises and papers.

Every participant should prepare by reading the chapter(s) and attempt to solve the suggested
exercises.

The purpose is to give an elaborate introduction to the topic of Differential Privacy in several
computer science contexts as well as practice working with related problems.

Reading list:
The book The Algorithmic Foundations of Differential Privacy by Cynthia Dwork and Aaron Roth.
Additional papers found by the lecturers, which concern topics presented in class.

Programme:
The following is a list of the lectures/exercise sessions and what should be prepared for each
session. Each lecture/exercise session will be 90 minutes.
Additional papers of particular interest may be added to the reading list during the seminar.

Date Topic/chapters
22/10/18 Chapters 1 & 2 - The Promise of Differential Privacy & Basics Terms
29/10/18 Chapter 3 - Basic Techniques and Composition Theorems
05/11/18 Chapter 4 - Releasing Linear Queries with Correlated Error
12/11/18 Chapter 5 - Generalizations
19/11/18 Chapter 6 - Boosting for Queries
26/11/18 - No lecture -
03/12/18 Chapters 7 & 8 - When worst Case Sensitivity is Atypical & Lower
Bounds and Separation Results
10/12/18 Chapter 9 - Differential Privacy and Computational Complexity
17/12/18 Chapter 10 - Differential Privacy and Mechanism Design
Christmas Break -
07/01/19 Chapter 11 - Differential Privacy and Machine Learning
14/01/19 Chapters 12 & 13 - Additional Models & Reflections
- Project work
01/03/19 Project Hand-in

The final project must be based on a paper or a topic not covered (or only briefly discussed) in class.

Prerequisites:
Basic knowledge of algorithms and privacy. The book gives an introduction to the topic.

Credits:
5 ECTS

Amount of hours the student is expected to use on the course:
Participation: 15 hours
Preparation: 50 hours
Project work: 70 hours
Total: 135 hours

Prerequisites:
Basic knowledge of algorithms and privacy. The book gives an introduction to the topic.

Participants:
Maximum 10 participants.