A Reductions Approach to Fair Classification
It provides a systematic method for enforcing fairness constraints in classification, addressing bias issues for affected groups, though it is incremental in building on existing definitions.
The paper tackles fair binary classification by reducing it to cost-sensitive classification problems, achieving lower empirical error than prior baselines across various datasets.
We present a systematic approach for achieving fairness in a binary classification setting. While we focus on two well-known quantitative definitions of fairness, our approach encompasses many other previously studied definitions as special cases. The key idea is to reduce fair classification to a sequence of cost-sensitive classification problems, whose solutions yield a randomized classifier with the lowest (empirical) error subject to the desired constraints. We introduce two reductions that work for any representation of the cost-sensitive classifier and compare favorably to prior baselines on a variety of data sets, while overcoming several of their disadvantages.