This course will study the sources and measures of unfairness stemming from machine learning, as well as possible interventions to alleviate said unfairness. The class will contain a mix of assignments: there will be 5 assignments. Each assignment will contain 3 parts: a programming task, several mathematical questions which will require rigorous proofs to answer, and an analysis/writing component.
This course is open to graduate students with background in machine learning and algorithm design. Any undergraduate who wishes to take the course must ask for the instructor’s permission.
This class will cover several different perspectives on fairness in machine learning. Topics will include:
All class announcements will be posted on Piazza.
Please read the assigned readings before attending lecture and lab.
Fall 2018 schedule
|Week 1 Aug 20|
|Lecture 1 (9/20, Mon):
|No Class (9/22, Wed)|
|Week 2 Aug 27|
|Lecture 2 (8/27, Mon)
What assumptions do we make about our world, our data, our goals?
| On the (Im)Possibility of Fairness
Fairness through Awareness
| Lecture 3, (8/29, Wed)
Legalese and (some very limited) historical context.
Big Data's Disparate Impact
(Optional) The Sociology of Discrimination: Racial Discrimination in Employment, Housing, Credit, and Consumer Markets
|Week 3 Sept 3|
| No Class (9/3, Mon)
| Lecture 4, (9/5, Wed)
Fairness through Awareness,
(Optional) Fairness in Learning: Classic and Contextual Bandits
|Week 4 Sept 10|
|No Class (9/10, Mon)|
| Lecture 5, (9/12, Wed)
Group Fairness Measures: An introduction.
Equality of Opportunity in Supervised Learning
(Optional/Recommended) Fair Prediction with Disparate Impact: A Study of Bias in Recidivism Prediction Instruments
(Optional/Recommended) Trade-Offs in the Fair Determination of Risk Scores
|Week 5 Sept 17|
|Lecture 6 (9/17, Mon)
: Continuation of Group Fairness Measures.
|No Class, (9/19, Wed)|
|Week 6 Sept 24|
|Lecture 7 (9/24, Mon)
Interpolating between Individual and Group Fairness Measures
|Preventing Fairness Gerrymandering|
| Lecture 8, (9/26, Wed):
A critique of these ``fairness" constraints
|Algorithmic decision making and the cost of fairness|
|Week 7 Oct 1|
|Lecture 9 (10/1, Mon):
What's wrong with classification, anyway?
|Invisible Mediators of Action: Classification and the Ubiquity of Standards|
| Lecture 10, (10/3, Wed):
| Learning Fair Representations
Optional, but highly recommended: Learning Adversarially Fair and Transferable Representations
|Week 8 Oct 8|
| No Class (10/8, Mon)
| Lecture 11, (10/11, Wed):
Is this data (un)fair? How should we "fix" it?
| Raw Data is an Oxymoron
And a recap of (Optional) The Sociology of Discrimination: Racial Discrimination in Employment, Housing, Credit, and Consumer Markets
and Data preprocessing techniques for classification without discrimination
|Week 9 Oct 15|
| Lecture 12 (10/15, Mon)
: Economics and Discrimination 1
| Lecture 13, (10/18, Wed):
: Economics and Discrimination 2
|Week 10 Oct 22|
| Lecture 14, (10/22, Mon):
| Lecture 15, (10/24, Wed):
|Week 11 Oct 29|
| Lecture 16, (10/29, Mon):
| Lecture 17, (10/31, Wed):
|Week 12 Nov 5|
| Lecture 18 (11/5, Mon):
|Lecture 19, (11/7, Wed)|
|Week 13 Nov 12|
| Lecture 20 (11/12, Mon):
|Lecture 21, (11/15, Wed)|
|Week 14 Nov 19|
| Lecture 22 (11/12, Mon):
|No Class, Student Recess, (11/19, Wed)|
|Week 15 Nov 26|
| Lecture 23 (11/26, Mon):
|Lecture 24, (11/28, Wed)|
|Week 16 (12/3, NIPS week)|
| Perhaps no class? (12/3, Mon)
|No Class, Reading Period (12/5, Wed)|
|Week 17 (12/10, Final Exams)|