Preference Informed Fairness
summary
Proposes preference-informed individual fairness: classifiers may deviate from individual-fair outcomes only when each affected person prefers the deviation to a suitable fictitious individually-fair reference allocation. Formalizes preferences via expected-utility and shows PIIF is a relaxation of individual fairness and envy-freeness. Gives an algorithmic formulation as a convex program (introducing pairwise fictitious-allocation variables, roughly O(n^2) extra variables) and argues the constraints are linear so the optimization is tractable under standard assumptions.
AI-written summary · about
chapters (5)
related talks
31mIndividual Fairness in Hindsight
29mIndividual Fairness in Advertising Auctions through Inverse Proportionality
46mEliciting and Enforcing Subjective Individual Fairness
53mGroups and Individuals in Fairness and Probability
37mAlgorithmic Fairness From The Lens Of Causality And Information Theory
44mProcedural Justice in Algorithmic Fairness: Transparency, Control and Participation Perspectives
14mFair And Reliable Machine Learning For High-Stakes Applications:approaches Using Information Theory
34mInterpretability and Algorithmic Fairness