HS2022: 62116 Fairness and Privacy in Machine Learning

This course explains the basic concepts of algorithmic privacy and fairness. The main application area is statistics and machine learning.

General Information

Course description
This course gives a thorough technical introduction to algorithmic privacy and fairness. The course work is centered around individual assignments and group project work.

Assessment: 80% project work, 20% exam
Syllabus
1. Algorithmic privacy, fairness and reproducibility
2. Privacy and anonymity
3. Differential privacy
4. Approximate differential privacy
5. Privacy amplification
6. Group fairness: Equalised odds
7. Group fairness: Balance and calibration
8. Individual fairness: Meritocracy
9. Individual fairness: Smoothness
10. Reproducibility
Target Group
Highly-motivated master students interested in algorithmic privacy and fairness.

Prerequisites, in order of importance:
1. Elementary Probability.
2. Good programming skills.
3. Multivariate calculus
4. Linear algebra

The advanced seminar course explores further topics in the area in more detail. It is possible to take both courses in parallel.

Lernziele und Links

Links
https://github.com/olethrosdc/ml-society-science

General

Language
English
Copyright
All rights reserved

Contact

Name
Christos Dimitrakakis
E-Mail
christos.dimitrakakis@unine.ch
Consultation
By appointment

Availability

Access
Unlimited
Admittance
You can join this course directly.
Registration Period
Unlimited
Period of Event
27. Sep 2022 - 16. Dec 2022

Personal Data Visible to Course Administrators

Data Types of the Personal Profile
Username
First Name
Last Name
E-Mail