Cardinal objectives serve as intuitive targets in fair machine learning by summarizing utility
(welfare) or disutility (malfare) $ u $ over $ g $ groups. Under standard axioms, all welfare …

In fair machine learning, one source of performance disparities between groups is overfitting
to groups with relatively few training samples. We derive group-specific bounds on the …

Welfare measures overall utility across a population, whereas malfare measures overall
disutility, and the social planner's problem can be cast either as maximizing the former or …

The original position or veil of ignorance argument of John Rawls, perhaps the most famous
argument for egalitarianism, states that our concept of fairness, justice, or welfare should be …

Defining demographic categories for fair learning frequently requires both private
information collection at the individual level and expert judgment in the construction of …

We propose a welfare-centric fair reinforcement-learning setting, in which an agent enjoys
vector-valued reward from a set of beneficiaries. Given a welfare function W (·), the task is to …

Introduction My research has always been broadly interdisciplinary, with a focus on
introducing rigorous modern statistical techniques and finite-sample concentration-of …