Course syllabus

Jump to today

DAT655 / DIT117 Data Privacy lp2 HT25 (7.5 hp)

Course is offered by the Department of Computer Science and Engineering

Contact Details

Examiner and Lecturer

  • Prof. Alejandro Russo
  • Email: russo@chalmers.se
  • Faculty profile
  • Office hours: By appointment (check schedule)

Teaching Assistants

  • Alex Ionescu
  • Email: ionescua@chalmers.se
  • Faculty profile
  • Office hours: Usually on Fridays 13:15-15:00 (check schedule), Room X

Course Representatives

Program Email Name of student
MPCSN <TBA> <TBA>
MPCSN <TBA> <TBA>
MPCSC <TBA> <TBA>
MPCSC <TBA> <TBA>
MPCSC <TBA> <TBA>

Course Purpose

This course provides students with theoretical foundations and practical skills in data privacy, with a particular focus on differential privacy mechanisms and privacy-preserving data analysis. Students will learn to identify privacy risks in data-sharing scenarios, understand the principles behind privacy regulations like GDPR, and develop systems that implement differential privacy techniques. The course covers both theoretical aspects (privacy definitions, composition theorems, sensitivity analysis) and practical implementation (building differentially private systems, testing mechanisms, generating synthetic data). Through programming assignments and examination, students will gain hands-on experience in designing and evaluating privacy-preserving systems.

Schedule

Schedule is available on TimeEdit - search for course code DAT655 or DIT117.

Course Literature

Mandatory Literature

The course is organized around two free online books:

  • The Algorithmic Foundations of Differential Privacy

    • Authors: Cynthia Dwork and Aaron Roth
    • Access: Free PDF
    • Coverage: Theoretical foundations, composition theorems, sensitivity analysis

  • Programming Differential Privacy

    • Authors: Joseph P. Near and Chiké Abuah
    • Access: Free online book
    • Coverage: Practical implementation, systems design, real-world applications

Additional Materials

The course also includes several research papers indicated in each lecture. These papers are provided through Canvas or direct links in the lecture materials.

Course Design

Learning Activities

The course combines theoretical instruction with practical programming assignments:

Lectures

  • Theoretical foundations of data privacy and differential privacy
  • Privacy regulations (GDPR) and their implications
  • De-identification techniques and their vulnerabilities
  • Differential privacy mechanisms (Laplace, exponential, etc.)
  • Composition theorems and privacy budget management
  • Sensitivity analysis and stability tracking
  • Testing differential privacy implementations
  • Floating-point attacks on privacy systems

Programming Assignments (Labs)

Assignment 1: Reconstruction Attack (Due: November 19, 2026)

  • Implement the Dinur-Nissim reconstruction attack
  • Understand vulnerabilities in noisy database systems
  • Available in both Haskell and Python
  • Work in pairs or individually/group of three with explicit authorization of the lecturer
  • Repository: dataprivacy-course on GitHub

Assignment 2: Building a Differential Privacy DSL (Due: December 8, 2026)

  • Design and implement a modular differential privacy library
  • Implement budget tracking and composition
  • Build data transformations with stability tracking
  • Implement Laplace mechanism for count and sum queries
  • Test statistical soundness using \((\alpha,\beta)\)-accuracy tests
  • Available in both Haskell and Python
  • Work in pairs or individually with permission

Supervision and Feedback

  • Teaching assistants provide feedback within three working days after submission
  • Office hours available for consultation (see contact details above)
  • Questions can be submitted via Canvas or email

Digital Tools and Communication

  • Canvas: Primary learning platform for course materials, announcements, assignment submissions
  • Email: Use Canvas messages system
  • GitHub: Lab materials and skeleton code

Assignment Submission and Revisions

Important Policies:

  • Hard Deadlines: Deadlines are strictly enforced
  • Extension Requests: Contact instructors BEFORE the deadline with a valid reason and realistic new deadline proposal
  • Submission Requirements: You must submit a serious attempt by the initial deadline
  • Final Deadlines: One revision is allowed, must be submitted before the final deadline
Assignment Initial Deadline Final Deadline
1 November 19, 2026 December 1, 2026
2 December 8, 2026 December 17, 2026

If you fail to submit by the final deadline, your submission will be rejected and you must wait for the re-examination period.

Academic Integrity

Cheating is taken very seriously. Cases suspected of cheating will be reported to the Disciplinary Committee for further investigation. This can lead to exclusion from lectures, computer rooms, and exams.

Legitimate Cooperation:

  • You may orally discuss exercises and assignments with other students
  • Programming assignments may be completed in pairs (size two only)
  • Once you cooperate with someone on an assignment, you must submit together with that person
  • You may NOT share code by any means (email, printing, files, phone, web, GitHub, etc.)
  • Both giving and receiving code are equally serious violations

Any exceptions must be discussed with the lecturer immediately.

Changes Made Since Last Occasion

This is the 2026/2027 iteration of the course. Specific changes will be communicated at the course start.

Learning Objectives and Syllabus

Learning Objectives

Knowledge and Understanding:

  • Summarize the principles behind privacy laws like GDPR
  • Identify data-sharing challenges with a focus on preserving privacy, including anonymization techniques and their vulnerabilities
  • Explain Differential Privacy (DP) theoretical foundations and their properties
  • Develop practical systems and tools for implementing DP

Skills and Abilities:

  • Critically analyze privacy risks in data-sharing scenarios and evaluate the strengths and weaknesses of anonymization techniques and DP
  • Design and develop systems employing DP techniques
  • Test DP mechanisms and analyze their statistical soundness using established methods

Judgement Ability and Approach:

  • Assess privacy-preserving systems and frameworks, considering the trade-offs between utility and privacy
  • Evaluate the choice of privacy-preserving techniques based on the context, considering both the theoretical guarantees and practical limitations of mechanisms like k-anonymity and differential privacy
  • Discern the ethical implications of data privacy, and align solutions with legal frameworks like the GDPR, helping with compliance in real-world scenarios

Examination Form

Examination Components

The course examination consists of:

  1. Programming Assignments (Pass/Fail) - 2 assignments - Assignment 1: Reconstruction Attack - Assignment 2: Differential Privacy DSL - Requirement: Both assignments must be passed to be eligible for the final exam

  2. Written Examination (Grades: 3, 4, 5) - Format: Written exam - Final Grade: The course grade is determined by the exam grade

Assignment Grading

  • Grading Scale: Fail (U) or Pass (G)
  • Evaluation Criteria:
  • Correctness of implementation
  • Code quality and best practices
  • Working examples demonstrating proper usage
  • Passing provided test suites
  • Understanding of privacy concepts demonstrated in code

Examination Details

Written Exam:

  • Form: Written examination (format TBA)
  • Aids: Will be specified before the exam
  • Grading: Swedish scale (3, 4, 5)
  • Time and Place: Will be announced via Canvas and TimeEdit

Assignment Submissions:

  • Format: Code submission via Canvas
  • Compulsory Elements:
  • Complete implementation of required functions
  • Working examples
  • Passing test suites
  • Templates: Skeleton code provided in GitHub repository
  • Deadline Policy: See "Assignment Submission and Revisions" section above

Requirements to Pass

  1. Pass both programming assignments (Assignment 1 and Assignment 2)
  2. Pass the written examination (grade 3 or higher)

Re-examination

Students who fail the regular examination have the right to re-examination according to Chalmers/GU regulations.

Course Summary

Course Modules:

The course is organized into the following modules (delivered as PDF documents and lectures):

  1. Course Introduction and Organization
  2. Medallion Architecture for Privacy
  3. Data Privacy Fundamentals
  4. De-identification Techniques (input focus)
  5. De-identification Techniques (output focus)
  6. Differential Privacy Preliminaries
  7. Introduction to Differential Privacy
  8. Laplace Mechanism
  9. Differential Privacy Properties
  10. Sensitivity Analysis
  11. Building Statistical Databases with DP
  12. Floating Point Attacks
  13. Testing Differential Privacy

Key Dates:

Date Event Details
November 19, 2026 Assignment 1 Deadline Reconstruction Attack
December 1, 2026 Assignment 1 Final Deadline Last chance for revision
December 8, 2026 Assignment 2 Deadline DP DSL Implementation
December 17, 2026 Assignment 2 Final Deadline Last chance for revision
TBA Written Exam Check TimeEdit for date/time/location

For questions about the course, contact the lecturer or teaching assistants via email or Canvas.

Course summary:

Course Summary
Date Details Due