On the Impossibility of non-Trivial Accuracy in Presence of Fairness Constraints

Carlos Pinzón (1) , Catuscia Palamidessi (1) , Pablo Piantanida (2, 3) , Frank Valencia (1, 3)

1 COMETE - Concurrency, Mobility and Transactions
2 CentraleSupélec
3 CNRS - Centre National de la Recherche Scientifique

One of the main concerns about fairness in machine learning (ML) is that, in order to achieve it, one may have to trade off some accuracy. To overcome this issue, Hardt et al. proposed the notion of equality of opportunity (EO), which is compatible with maximal accuracy when the target label is deterministic with respect to the input features. In the probabilistic case, however, the issue is more complicated: It has been shown that under differential privacy constraints, there are data sources for which EO can only be achieved at the total detriment of accuracy, in the sense that a classifier that satisfies EO cannot be more accurate than a trivial (i.e., constant) classifier. In our paper we strengthen this result by removing the privacy constraint. Namely, we show that for certain data sources, the most accurate classifier that satisfies EO is a trivial classifier. Furthermore, we study the trade-off between accuracy and EO loss (opportunity difference), and provide a sufficient condition on the data source under which EO and non-trivial accuracy are compatible.

Domaines: Intelligence artificielle [cs.AI]; Apprentissage [cs.LG]

Carlos Pinzón, Catuscia Palamidessi, Pablo Piantanida, Frank Valencia. On the Impossibility of non-Trivial Accuracy in Presence of Fairness Constraints. Proceedings of the AAAI 36th Conference on Artificial Intelligence, Feb 2022, Vancouver / Virtual, Canada. pp.7993-8000, ⟨10.1609/aaai.v36i7.20770⟩. ⟨hal-03452324⟩