When: The workshop will be held on June 29th 8:45-11:45 AM

Speakers: Cynthia Dwork, Michael Kearns, and Toni Pitassi

Overview. The increasing reach of algorithmic classification and decision making into our daily lives — from advertising to candidate filtering for jobs to incarceration — has given rise to an explosion of research into the ethics embodied by these algorithms; in a word, are they “fair”? But what is fairness? Can we test for it? Can we achieve it? Are there limits? This tutorial covers theoretical advances and approaches in machine learning.

Rough outline and preliminary schedule. The tutorial will cover the following topics:

• 8:45-9:15 (Dwork): Fundamentals: Definitions of fairness, incompatibility of some fairness notions, composition of individually fair classifiers
• 9:20- 10:10 (Kearns): Algorithms and hardness results: fairness in learning (bandit settings), multicalibration, learning with an unknown metric, multifairness, anti-gerrymandering

10:15-10:25: Break

• 10:25-10:45 (Dwork): Composition results for group fairness
• 10:50-11:40 (Pitassi): Fairness in Machine Learning: fair representations and fair transfer learning

Organizers: Cynthia Dwork and Guy Rothblum