Jeff EdmondsAProfessor
Electrical Engineering and Computer Science Department
Electrical Engineering and Computer Science Department
Member, IC@L
Member, Centre for Vision Research
2021 – 2022 Research Highlights
Ensuring Fairness Despite Differences in Environment:
If a disadvantage person performs as well as an advantaged person, despite being disadvantaged, then we might assume she has some underling talent which we want in the people that we hire. We model this by having the talent of each person be chosen from the same distribution regardless of race, gender, or the colour of their skin. However, advantaged, and disadvantaged people have their environment chosen very different distributions. Their performance is assumed to be the sum of this talent and this environment. Conditioned on the two people having the same performance, we prove that for most distributions, the disadvantage person is likely to be more talented than the advantaged person. Seems obvious, but considering arbitrary distributions made the math harder.