Recipe: Algorithmic Fairness
Creators: Somaieh Nikpoor, Willie Costello
Objective: This list will get you started with concepts and practices in algorithmic fairness
Audience Level: Beginner
Main Concept: Algorithmic Bias and Fairness is the main concept you will learn through the following resources
Background Concept: You need to know bias in data and fairness definitions in order to learn the main concept
Subsequent Concept: Once you know the main concept you can learn fairness enhancing mechanisms
Resources
You should go through the following resources in the order that is provided:
- Getting Specific about Algorithmic Bias
Type: Video; Theory, Main Concept
Estimated time commitment: 30 mins
Why is this a good resource: It is a lecture by University of San Francisco professor and director of the USF Center for Applied Data Ethics that teaches you what algorithmic bias means.
How to use this resource: Watch the whole video (optional: there are other related resources from Rachel Thomas that you can read)
Instructor: Rachel Thomas
Link: https://www.youtube.com/watch?v=K7i_tnflZ64
2. 21 Fairness Definition and Their Politics
Type: Video Tutorial; Theory, Background Concept
Estimated time commitment: 60 mins
Why is this a good resource: It is a lecture by Princeton Professor, Arvind Narayanan that provides the basis you need to understand fairness in algorithm
How to use this resource: Watch the whole video (optional: there are other related resources from Arvind Narayanan that you can read)
Speaker: Arvind Narayanan
Link: https://www.youtube.com/embed/jIXIuYdnyyk
3. Social Data: Biases, Methodological Pitfalls, and Ethical Boundaries
Type: Paper; Theory, Background Concept
Estimated time commitment: 20 mins
Why is this a good resource: You learn about different type of data biases and how to measure them
How to use this resource: Read section 3 to learn about types of biases in data (optional: you can read other sections of the paper as well as the resources recommended on section 10.3)
Author: Alexandra Olteanu, Carlos Castillo, Fernando Diaz, Emre Kiciman
Link: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2886526
4. Dealing with Bias and Fairness in Building Data Science/ML/AI Systems
Type: GitHub Repo; Implementation, Subsequent Concept
Estimated time commitment: 90 mins
Why is this a good resource: Provides hand-on training to address bias and implement fairness in ML system
How to use this resource: watch the tutorial video and play with the code (optional: you can read the paper and other case studies)
Creator: Pedro Saleiro, Kit T. Rodolfa, Raid Ghani