Algorithmic Bias (AS.110.365)
Course Information
(Spring 2022)
Professor: Susama Agarwala
e-mail: susama.agarwala@jhuapl.edu
Office hours: Thursday 1:30-2:30. Virtual. Zoom meeting id: 160 1929 5670, Passcode: 624779
TA: Nandan Kulkarni
e-mail: nkulkar8@jhu.edu
Office hours: Tuesday 1-2. Room Kreiger 204
Meeting times:
Lecture: MW 3-4:15 Hodson 311
Section: F 3 -3:50 Hodson 305
Overly ambitious Course Syllabus
This is a superset of the material we will cover in class. As the course evolves, and we inevitably have to cut material, check back to see updated versions.
![](https://www.google.com/images/icons/product/drive-32.png)
Material covered
This section will be updated regularly with a few suggested links for the topics to be covered in class, as well as my notes from the material that I just covered. Click on the link associated to the lecture for my notes.
The supplementary readings are by no means the ONLY source of informations. Students are encouraged to use other statistics texts and sources that they are more comforatable with in order to get a different take on the material.
January 24: Introduction to AI Bias
January 26: Gauss Markov Assumptions and Linear Regression
Supplementary readings: Econometric Analysis (Greene) Chapters 2, 3, 4
Gauss-Markov Theorem and Ordinary Least Square Assumptions
You Tube lecture on Gauss-Markov
January 31: Gauss Markov Assumptions pt. 2 and ommitting confounding variables
Ommitting confounding variables
February 2: Confounding Variables and Pearson's R
Vaccination and political leaning picture taken from here
February 4: Section
February 7: More Pearson's R; Regression without Gauss Markov
Mostly Harmless Econometrics (Chapter 3)
February 9: Conditional Expectation Functions (Regression without Gauss Markov)
Mostly Harmless Econometrics (Chapter 3)
Lecture notes (note that proofs not done in class are available in these notes)
Feb 14: Review of CEF and regression interpretation concepts
Feb 16: Logistic Regressions
Latent variables for binary response
Econometric Analysis (Greene) Chapter 21
Feb 21 & 23:
Working with data scientifically
Hans Rosling:
1) Debunking third-world myths
3) How not to be ignorant about the world
What not to do with data
Cathy O'Neil:
1) Era of blind faith in big data must end
2) Weapons of Math Destruction
Joy Buolamwini
1) Compassion through computation
Timnit Gebru
1) How to stop AI from marginalizing communities
Ruha Benjamin
1) From park bench to lab bench (discussing research design more generally)
Feb 28:
Why binary predictors are tricky.R
March 2: Confusion matrices and ROC curves
March 7: Brief introduction to machine learning
Perceptrons:
Neural Networks:
Slightly more indepth article (with use cases)
Hands on :
March 9:
March 14: Review of homework 6
March 16:
March 28: Different measures of fairness
Fairness in Criminal Justice Risk Assessments: The State of the Art
March 30: Fairness Impossibility proofs
April 4: Efficiency/ Fairness Tradeoff
Equality of Opportunity in Supervised Learning
April 6: Efficiency/ Fairness Tradeoff
April 11: Discussion of Fairness and Tradeoffs
April 16: Discussion for HW 9
April 18: Algorithmic Redlining
April 21: Algorithmic Relining/ When tradeoffs aren't so bad
April 25: When tradeoffs aren't so bad
Aprli 27: Final Project
The Use and Misuse of Counterfactuals in Ethical Machine Learning
Chasing Your Long Tails: Differentially Private Prediction in Health Care Settings
“This Whole Thing Smacks of Gender”: Algorithmic Exclusion in Bioimpedance-based Body
Removing Spurious Features can Hurt Accuracy and Affect Groups Disproportionately
Algorithmic Fairness in Predicting Opiod Use Disorder Using Machine Learning
An Agent-based Model to Evaluate Interventions on Online Dating Platforms to Decrease
Homeworks
There will be 10 homework assignments this semester, posted on Fridays, and due the Monday 10 days following. Both assignments and solutions will be posted here.
Homework 1 (Due February 7)
Homework 2 (Due February 14)
Baseball data, source from here
Homework 4 (Due February 28)
Homework 5 (Due March 7)
Homework 6 (Due March 14)
Homework 7 (Due March 28)
Homework 8 (Due April 4)
Homework 9 (Due April 13)
Homework 10 (Due April 22)
Final Project (Due April 27)
Pick 2 papers from FAccT 2021, give a 20 minute presentation on the two combined.
Paper selection (Due April 15)
At least one paper must be on the topic of fairness
At least one paper must discuss a classifier or a regression method