Adversarial Fairness in the Labor Market
summary
Claims a production method called Adversarial Fairness that trains a KPI predictor against an adversary that tries to infer EEOC protected classes, optimizing to a Nash equilibrium so the adversary becomes degenerate. Uses GAN-style objective plus standard predictors (logistic, random forest, neural nets), measures remaining bias with the AUC of the adversary rather than the 80% rule, and reports deployment on 500,000 applicants with class distributions rendered effectively indistinguishable and improved hiring ROI.
AI-written summary · about
chapters (9)
- 0:00Intro
- 1:43ABSTRACT SUMMARY
- 2:58INEFFICIENCY IN THE LABOR MARKET
- 6:07MACHINE LEARNING TEMPLATE
- 11:35BIAS MITIGATION IN AI
- 13:12ADVERSARIAL FAIRNESS
- 14:37TRAINING THE GAN
- 15:25p-rule IMPACT
- 16:01KEY TAKE-AWAYS
related talks
01h 30mFairness Tales: How To Measure And Mitigate Unfair Bias in Machine Learning Models
47mEthical Machine Learning
13mCan AI design a fairer future?
42mFairness in Machine Learning: From Theory to Practice
34mInterpretability and Algorithmic Fairness
11mAddressing Privacy, Bias, and Fairness in AI Systems
01h 31mHow To Measure And Mitigate Unfair Bias in Machine Learning Models
38mRepresentation via Representations: Improving Generalization to Unseen Populations